{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T14:40:29Z","timestamp":1788273629594,"version":"build-2803163510"},"update-to":[{"DOI":"10.12688\/wellcomeopenres.24189.1","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T00:00:00Z","timestamp":1783900800000}}],"reference-count":109,"publisher":"F1000 Research Ltd","license":[{"start":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T00:00:00Z","timestamp":1783900800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100010269","name":"Wellcome Trust","doi-asserted-by":"publisher","award":["226945"],"award-info":[{"award-number":["226945"]}],"id":[{"id":"10.13039\/100010269","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["wellcomeopenresearch.org"],"crossmark-restriction":false},"short-container-title":["Wellcome Open Res"],"abstract":"<ns5:p>\n                    Background Risk of new episodes is a constant concern for people living with bipolar disorder (BD). Research into sleep and circadian rhythm disruption (SCRD) suggests that relapse risk may be measurable in 24-hour rest-activity rhythms (RAR) via actigraphy. The primary aim of the Tipping Point project is to develop an algorithm forecasting relapse based on RAR. A secondary aim is to explore SCRD mechanisms linking RAR to mood dynamics in BD. The mechanistic investigation involves analysis of biological correlates of RAR, and exploration of system instability as a predictor of two types of state transition (relapse and recovery). Method Three prospective studies will enrol adults living with BD I or II. Study 1 (Australia) recruits\n                    <ns5:italic>N<\/ns5:italic>\n                    \u00a0=\u00a0100 individuals with interepisode BD who will undergo three contiguous 26-week monitoring epochs. Each epoch commences with 14\u00a0days of actigraphy, followed by retrospective interviews at 13 and 26\u00a0weeks. RAR predictors of time to first relapse will be explored using three computational frameworks - statistical risk, probabilistic network graphs and non-linear dynamic models. Study 2 (India) replicates one of the 26-week epochs of Study 1, recruiting 25 individuals with BD and 25 matched healthy controls. Study 2 investigates the replicability of models identified in Study 1 and tests the mechanistic association between amplitude of the RAR and amplitude of clock gene expression. Study 3 (New Zealand) explores system instability as a mechanism, by investigating whether complex system indicators of relapse risk (from Study 1) also forecast recovery from acute episodes. In Study 3, actigraphy data will be collected from people experiencing a severe episode of BD depression (\n                    <ns5:italic>N<\/ns5:italic>\n                    \u00a0=\u00a030) or mania (\n                    <ns5:italic>N<\/ns5:italic>\n                    \u00a0=\u00a015). Discussion The Tipping Point project uses a multi-study design to develop a mechanistically-informed predictive algorithm for relapse risk in BD.\n                  <\/ns5:p>","DOI":"10.12688\/wellcomeopenres.24189.2","type":"journal-article","created":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T10:39:06Z","timestamp":1783939146000},"page":"320","update-policy":"https:\/\/doi.org\/10.12688\/wellcomeopenres.crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Forecasting the course of bipolar disorder using rest-activity rhythms: Protocol for a multi-study modelling project"],"prefix":"10.12688","volume":"10","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7208-5603","authenticated-orcid":false,"given":"Greg","family":"Murray","sequence":"first","affiliation":[{"name":"Centre for Mental Health and Brain Sciences Melbourne Victoria Swinburne University of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9960-5768","authenticated-orcid":false,"given":"Sandipan","family":"Ray","sequence":"additional","affiliation":[{"name":"Department of Biotechnology Hyderabad Indian Institute of Technology Hyderabad"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8695-3966","authenticated-orcid":false,"given":"Richard","family":"Porter","sequence":"additional","affiliation":[{"name":"Mental Health Clinical Research Unit Christchurch Otago University of Otago"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3870-798X","authenticated-orcid":false,"given":"Fatemeh","family":"Hadaeghi","sequence":"additional","affiliation":[{"name":"Institute of Computational Neuroscience Hamburg Hamburg Universitatsklinikum Hamburg-Eppendorf"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9902-0858","authenticated-orcid":false,"given":"Denny","family":"Meyer","sequence":"additional","affiliation":[{"name":"Centre for Mental Health and Brain Sciences Melbourne Victoria Swinburne University of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maree","family":"Choi","sequence":"additional","affiliation":[{"name":"Centre for Mental Health and Brain Sciences Melbourne Victoria Swinburne University of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-4474-4649","authenticated-orcid":false,"given":"Sara","family":"Lapsley","sequence":"additional","affiliation":[{"name":"Department of Psychology Ottawa Carleton University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Matthew","family":"Tennant","sequence":"additional","affiliation":[{"name":"Mental Health Clinical Research Unit Christchurch Otago University of Otago"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bridgette","family":"Thwaites","sequence":"additional","affiliation":[{"name":"Mental Health Clinical Research Unit Christchurch Otago University of Otago"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-7801-5577","authenticated-orcid":false,"given":"Ly","family":"Nguyen","sequence":"additional","affiliation":[{"name":"Centre for Mental Health and Brain Sciences Melbourne Victoria Swinburne University of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hailey","family":"Tremain","sequence":"additional","affiliation":[{"name":"Centre for Mental Health and Brain Sciences Melbourne Victoria Swinburne University of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-7076-2067","authenticated-orcid":false,"given":"Saibal","family":"Saha","sequence":"additional","affiliation":[{"name":"Department of Biotechnology Hyderabad Indian Institute of Technology Hyderabad"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7203-8601","authenticated-orcid":false,"given":"Jan","family":"Scott","sequence":"additional","affiliation":[{"name":"Psychological Medicine Newcastle Newcastle University"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"2560","published-online":{"date-parts":[[2026,7,13]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"1575-1586","DOI":"10.1016\/S0140-6736(13)61611-6","article-title":"Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010.","volume":"382","author":"H Whiteford","year":"2013","journal-title":"Lancet."},{"key":"ref2","first-page":"379-386","article-title":"Medical care costs and hospitalization in patients with bipolar disorder treated with atypical antipsychotics.","volume":"5","author":"J Bergeson","year":"2012","journal-title":"Am Health Drug Benefits."},{"key":"ref3","doi-asserted-by":"publisher","first-page":"194-214","DOI":"10.1111\/bdi.12775","article-title":"Functional outcome assessment in bipolar disorder: A systematic literature review.","volume":"21","author":"M Chen","year":"2019","journal-title":"Bipolar Disord."},{"key":"ref5","doi-asserted-by":"publisher","first-page":"12","DOI":"10.1186\/s40345-024-00333-y","article-title":"Longitudinal studies of bipolar patients and their families: translating findings to advance individualized risk prediction, treatment and research.","volume":"12","author":"A Duffy","year":"2024","journal-title":"Int J Bipolar Disord."},{"key":"ref6","doi-asserted-by":"publisher","first-page":"129-137","DOI":"10.1016\/j.jpsychires.2020.12.041","article-title":"Clinical predictors of recurrences in bipolar disorders type 1 and 2: A FACE-BD longitudinal study.","volume":"134","author":"B Etain","year":"2021","journal-title":"J Psychiatr Res."},{"key":"ref7","doi-asserted-by":"publisher","first-page":"104552","DOI":"10.1016\/j.neubiorev.2022.104552","article-title":"Machine learning approaches for prediction of bipolar disorder based on biological, clinical and neuropsychological markers: A systematic review and meta-analysis.","volume":"135","author":"F Colombo","year":"2022","journal-title":"Neurosci Biobehav Rev."},{"key":"ref8","doi-asserted-by":"publisher","first-page":"434-447","DOI":"10.1111\/acps.13765","article-title":"Digital phenotyping in bipolar disorder: Using longitudinal Fitbit data and personalized machine learning to predict mood symptomatology.","volume":"151","author":"J Lipschitz","year":"2025","journal-title":"Acta Psychiatrica Scandinavica."},{"key":"ref9","doi-asserted-by":"publisher","first-page":"525","DOI":"10.1038\/s41398-021-01652-9","article-title":"Association between circadian activity rhythms and mood episode relapse in bipolar disorder: a 12-month prospective cohort study.","volume":"11","author":"Y Esaki","year":"2021","journal-title":"Translational Psychiatry."},{"key":"ref10","doi-asserted-by":"publisher","first-page":"e10075","DOI":"10.32872\/cpe.10075","article-title":"Building an Early Warning System for Depression: Rationale, Objectives, and Methods of the WARN-D Study.","volume":"5","author":"E Fried","year":"2023","journal-title":"Clin Psychol Eur."},{"key":"ref11","doi-asserted-by":"publisher","first-page":"150-151","DOI":"10.1002\/wps.21154","article-title":"Sleep and circadian rhythm disturbances: plausible pathways to major mental disorders?","volume":"23","author":"I Hickie","year":"2024","journal-title":"World Psychiatry."},{"key":"ref12","doi-asserted-by":"publisher","first-page":"e2214756121","DOI":"10.1073\/pnas.2214756121","article-title":"The sleep-circadian interface: A window into mental disorders.","volume":"121","author":"N Meyer","year":"2024","journal-title":"Proc Natl Acad Sci USA."},{"key":"ref13","doi-asserted-by":"publisher","first-page":"104585","DOI":"10.1016\/j.neubiorev.2022.104585","article-title":"A systematic review and meta-analysis of sleep and circadian rhythms disturbances in individuals at high-risk of developing or with early onset of bipolar disorders.","volume":"135","author":"J Scott","year":"2022","journal-title":"Neurosci Biobehav Rev."},{"key":"ref14","doi-asserted-by":"publisher","first-page":"2281-2289","DOI":"10.1017\/S0033291720004171","article-title":"An ecological study of objective rest-activity markers of lithium response in bipolar-I-disorder.","volume":"52","author":"J Scott","year":"2022","journal-title":"Psychol Med."},{"key":"ref15","doi-asserted-by":"publisher","first-page":"232-263","DOI":"10.1111\/bdi.13165","article-title":"Neurobiological and behavioral mechanisms of circadian rhythm disruption in bipolar disorder: A critical multi-disciplinary literature review and agenda for future research from the ISBD task force on chronobiology.","volume":"24","author":"M McCarthy","year":"2022","journal-title":"Bipolar Disord."},{"key":"ref16","doi-asserted-by":"publisher","first-page":"780726","DOI":"10.3389\/fpsyt.2022.780726","article-title":"Toward a Digital Future in Bipolar Disorder Assessment: A Systematic Review of Disruptions in the Rest-Activity Cycle as Measured by Actigraphy. Frontiers.","volume":"13","author":"P Panchal","year":"2022","journal-title":"Psychiatry."},{"key":"ref17","doi-asserted-by":"publisher","first-page":"37-41","DOI":"10.53841\/bpscpf.2014.1.253.37","article-title":"Using actigraphy to monitor sleep-wake patterns in bipolar disorder - A case study.","volume":"1","author":"B Bullock","year":"2014","journal-title":"Clin Psychol Forum."},{"key":"ref18","doi-asserted-by":"publisher","first-page":"468-478","DOI":"10.1111\/bdi.13430","article-title":"Early warning signals observed in motor activity preceding mood state change in bipolar disorder.","volume":"26","author":"P Jakobsen","year":"2024","journal-title":"Bipolar Disorders."},{"key":"ref61","doi-asserted-by":"publisher","first-page":"783-792","DOI":"10.1177\/0004867416642022","article-title":"Toward a complex system understanding of bipolar disorder: A chaotic model of abnormal circadian activity rhythms in euthymic bipolar disorder.","volume":"50","author":"F Hadaeghi","year":"2016","journal-title":"Aust N Z J Psychiatry."},{"key":"ref19","doi-asserted-by":"publisher","first-page":"20190098","DOI":"10.1098\/rsfs.2019.0098","article-title":"Sleep, circadian rhythms and health.","volume":"10","author":"R Foster","year":"2020","journal-title":"Interface Focus."},{"key":"ref20","doi-asserted-by":"publisher","first-page":"R128-r138","DOI":"10.1093\/hmg\/ddx240","article-title":"The genetics of circadian rhythms, sleep and health.","volume":"26","author":"A Jagannath","year":"2017","journal-title":"Hum Mol Genet."},{"key":"ref21","doi-asserted-by":"publisher","first-page":"164-179","DOI":"10.1038\/nrg.2016.150","article-title":"Transcriptional architecture of the mammalian circadian clock.","volume":"18","author":"J Takahashi","year":"2017","journal-title":"Nat Rev Genet."},{"key":"ref22","doi-asserted-by":"publisher","first-page":"e1007666","DOI":"10.1371\/journal.pcbi.1007666","article-title":"Methods detecting rhythmic gene expression are biologically relevant only for strong signal.","volume":"16","author":"D Laloum","year":"2020","journal-title":"PLoS Comput Biol."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"136-148","DOI":"10.1177\/0748730410395527","article-title":"Altered expression of circadian clock genes in human chronic myeloid leukemia.","volume":"26","author":"M Yang","year":"2011","journal-title":"J Biol Rhythms."},{"key":"ref24","doi-asserted-by":"publisher","first-page":"188-194","DOI":"10.1093\/sleep\/17.2.188","article-title":"Sleep as a Biological Rhythm: A Historical Review.","volume":"17","author":"W Webb","year":"1994","journal-title":"Sleep."},{"key":"ref25","doi-asserted-by":"publisher","first-page":"131-143","DOI":"10.1111\/jsr.12371","article-title":"The two-process model of sleep regulation: a reappraisal.","volume":"25","author":"A Borb\u00e9ly","year":"2016","journal-title":"J Sleep Res."},{"key":"ref26","doi-asserted-by":"publisher","first-page":"100080","DOI":"10.1016\/j.nbscr.2022.100080","article-title":"Human circadian rhythm studies: Practical guidelines for inclusion\/exclusion criteria and protocol.","volume":"13","author":"Y Yousefzadehfard","year":"2022","journal-title":"Neurobiol Sleep Circadian Rhythms."},{"key":"ref28","doi-asserted-by":"publisher","first-page":"741-773","DOI":"10.1111\/bdi.12847","article-title":"The chronotherapeutic treatment of bipolar disorders: A systematic review and practice recommendations from the ISBD task force on chronotherapy and chronobiology.","volume":"21","author":"J Gottlieb","year":"2019","journal-title":"Bipolar Disord."},{"key":"ref29","doi-asserted-by":"publisher","first-page":"49-65","DOI":"10.1038\/s41583-018-0088-y","article-title":"Rhythms of life: circadian disruption and brain disorders across the lifespan.","volume":"20","author":"R Logan","year":"2019","journal-title":"Nat Rev Neurosci."},{"key":"ref30","doi-asserted-by":"publisher","first-page":"507-514","DOI":"10.1016\/S2215-0366(18)30139-1","article-title":"Association of disrupted circadian rhythmicity with mood disorders, subjective wellbeing, and cognitive function: a cross-sectional study of 91 105 participants from the UK Biobank.","volume":"5","author":"L Lyall","year":"2018","journal-title":"Lancet Psychiatry."},{"key":"ref31","doi-asserted-by":"publisher","first-page":"29-36","DOI":"10.1097\/YPG.0000000000000215","article-title":"Missing a beat: assessment of circadian rhythm abnormalities in bipolar disorder in the genomic era.","volume":"29","author":"M McCarthy","year":"2019","journal-title":"Psychiatr Genet."},{"key":"ref32","doi-asserted-by":"publisher","first-page":"220-226","DOI":"10.1016\/j.jad.2022.03.007","article-title":"A network analysis of rest-activity rhythms in young people with emerging bipolar disorders.","volume":"305","author":"J Scott","year":"2022","journal-title":"J Affect Disord."},{"key":"ref33","doi-asserted-by":"publisher","first-page":"719-725","DOI":"10.1126\/science.8171325","article-title":"Mutagenesis and mapping of a mouse gene, Clock, essential for circadian behavior.","volume":"264","author":"M Vitaterna","year":"1994","journal-title":"Science."},{"key":"ref35","doi-asserted-by":"publisher","first-page":"637-654","DOI":"10.1080\/07853890.2018.1530449","article-title":"Prospects for circadian treatment of mood disorders.","volume":"50","author":"A H\u00fchne","year":"2018","journal-title":"Ann Med."},{"key":"ref36","doi-asserted-by":"publisher","first-page":"242-249","DOI":"10.1016\/j.biopsych.2013.02.019","article-title":"How might circadian rhythms control mood? Let me count the ways.","volume":"74","author":"C McClung","year":"2013","journal-title":"Biol Psychiatry."},{"key":"ref37","doi-asserted-by":"publisher","first-page":"143-146","DOI":"10.1192\/bjp.2023.150","article-title":"Sensitivity to light in bipolar disorder: implications for research and clinical practice.","volume":"224","author":"A Roguski","year":"2024","journal-title":"Br J Psychiatry."},{"key":"ref38","doi-asserted-by":"publisher","first-page":"689-704","DOI":"10.1016\/j.tips.2023.07.008","article-title":"Circadian biology to advance therapeutics for mood disorders.","volume":"44","author":"A Bhatnagar","year":"2023","journal-title":"Trends Pharmacol Sci."},{"key":"ref39","doi-asserted-by":"publisher","first-page":"86-96","DOI":"10.1177\/2167702613490158","article-title":"Reduced amplitude of the 24 hour activity rhythm: A biomarker of vulnerability to bipolar disorder?","volume":"2","author":"B Bullock","year":"2014","journal-title":"Clin Psychol Sci."},{"key":"ref40","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1186\/s40345-021-00220-w","article-title":"Evaluating endophenotypes for bipolar disorder.","volume":"9","author":"R Guglielmo","year":"2021","journal-title":"Int J Bipolar Disord."},{"key":"ref41","doi-asserted-by":"publisher","first-page":"351-354","DOI":"10.1097\/JCP.0000000000000699","article-title":"Comparison of the Effects of Quetiapine XR and Lithium Monotherapy on Actigraphy-Measured Circadian Parameters in Patients With Bipolar II Depression.","volume":"37","author":"J Hwang","year":"2017","journal-title":"J Clin Psychopharmacol."},{"key":"ref42","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1038\/s41746-023-00802-1","article-title":"40 years of actigraphy in sleep medicine and current state of the art algorithms.","volume":"6","author":"M Patterson","year":"2023","journal-title":"NPJ Digit Med."},{"key":"ref43","doi-asserted-by":"publisher","first-page":"870-879","DOI":"10.1111\/bdi.12021","article-title":"Evaluating sleep in bipolar disorder: comparison between actigraphy, polysomnography, and sleep diary.","volume":"14","author":"K Kaplan","year":"2012","journal-title":"Bipolar Disord."},{"key":"ref44","doi-asserted-by":"publisher","first-page":"46-58","DOI":"10.1016\/j.smrv.2014.06.006","article-title":"Sleep\u2013wake disturbance in interepisode bipolar disorder and high-risk individuals: A systematic review and meta-analysis.","volume":"20","author":"T Ng","year":"2015","journal-title":"Sleep Medicine Reviews."},{"key":"ref45","doi-asserted-by":"publisher","first-page":"947","DOI":"10.1176\/appi.books.9780890425596","year":"2013","journal-title":"Diagnostic and statistical manual of mental disorders: DSM-5."},{"key":"ref46","doi-asserted-by":"publisher","first-page":"189-196","DOI":"10.1001\/jamapsychiatry.2016.3459","article-title":"Activation in Bipolar Disorders: A Systematic Review.","volume":"74","author":"J Scott","year":"2017","journal-title":"JAMA Psychiatry."},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1093\/sleep\/zsad325","article-title":"State of the science and recommendations for using wearable technology in sleep and circadian research.","volume":"47","author":"M Zambotti","year":"2024","journal-title":"Sleep."},{"key":"ref49","doi-asserted-by":"publisher","first-page":"e816-e828","DOI":"10.1016\/S2589-7500(22)00152-2","article-title":"The promise of a model-based psychiatry: building computational models of mental ill health.","volume":"4","author":"T Hauser","year":"2022","journal-title":"Lancet Digit Health."},{"key":"ref50","doi-asserted-by":"publisher","first-page":"355","DOI":"10.1038\/s41746-024-01348-6","article-title":"The real-world association between digital markers of circadian disruption and mental health risks.","volume":"7","author":"M Lee","year":"2024","journal-title":"NPJ Digit Med."},{"key":"ref51","doi-asserted-by":"publisher","first-page":"290-307","DOI":"10.1016\/j.inffus.2019.04.001","article-title":"A survey on big data-driven digital phenotyping of mental health.","volume":"52","author":"Y Liang","year":"2019","journal-title":"Inform Fusion."},{"key":"ref52","doi-asserted-by":"publisher","first-page":"49-74","DOI":"10.1146\/annurev-clinpsy-102419-125032","article-title":"Personalized Models of Psychopathology.","volume":"16","author":"A Wright","year":"2020","journal-title":"Annu Rev Clin Psychol."},{"key":"ref53","doi-asserted-by":"publisher","first-page":"693-710","DOI":"10.1111\/bdi.12963","article-title":"Measuring circadian function in bipolar disorders: Empirical and conceptual review of physiological, actigraphic, and self-report approaches.","volume":"22","author":"G Murray","year":"2020","journal-title":"Bipolar Disord."},{"key":"ref54","doi-asserted-by":"publisher","first-page":"37","DOI":"10.1186\/s40345-020-00200-6","article-title":"An evidence map of actigraphy studies exploring longitudinal associations between rest-activity rhythms and course and outcome of bipolar disorders.","volume":"8","author":"J Scott","year":"2020","journal-title":"Int J Bipolar Disord."},{"key":"ref56","doi-asserted-by":"publisher","first-page":"350-367","DOI":"10.1093\/annweh\/wxaa007","article-title":"Actigraphy-Based Assessment of Sleep Parameters.","volume":"64","author":"D Fekedulegn","year":"2020","journal-title":"Ann Work Expo Health."},{"key":"ref55","doi-asserted-by":"publisher","first-page":"259-270","DOI":"10.1016\/0006-3223(95)00370-3","article-title":"Circadian rest-activity disturbances in Alzheimer\u2019s disease.","volume":"40","author":"E Van Someren","year":"1996","journal-title":"Biol Psychiatry."},{"key":"ref57","doi-asserted-by":"publisher","first-page":"117362","DOI":"10.1016\/j.eswa.2022.117362","article-title":"Mobile and wearable sensors for data-driven health monitoring system: State-of-the-art and future prospect.","volume":"202","author":"C Anikwe","year":"2022","journal-title":"Expert Syst Appl."},{"key":"ref58","doi-asserted-by":"publisher","first-page":"100602","DOI":"10.1016\/j.patter.2022.100602","article-title":"Modern views of machine learning for precision psychiatry.","volume":"3","author":"Z Chen","year":"2022","journal-title":"Patterns."},{"key":"ref59","doi-asserted-by":"publisher","first-page":"84","DOI":"10.1038\/s41746-023-00828-5","article-title":"Systematic review and meta-analysis of performance of wearable artificial intelligence in detecting and predicting depression.","volume":"6","author":"A Abd-Alrazaq","year":"2023","journal-title":"NPJ Digit Med."},{"key":"ref60","doi-asserted-by":"publisher","first-page":"115-124","DOI":"10.1016\/B978-0-12-821398-8.00001-1","author":"B Tietbohl-Santos","year":"2022","journal-title":"Biomarkers in Bipolar Disorders."},{"key":"ref62","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1186\/s40345-015-0043-z","article-title":"Circadian rhythmicity in emerging mood disorders: state or trait marker?","volume":"4","author":"A Grierson","year":"2016","journal-title":"Int J Bipolar Disord."},{"key":"ref63","doi-asserted-by":"publisher","first-page":"190-198","DOI":"10.1001\/jamapsychiatry.2018.3546","article-title":"Real-time Mobile Monitoring of the Dynamic Associations Among Motor Activity, Energy, Mood, and Sleep in Adults With Bipolar Disorder.","volume":"76","author":"K Merikangas","year":"2019","journal-title":"JAMA Psychiatry."},{"key":"ref64","doi-asserted-by":"publisher","first-page":"e1211-e1211","DOI":"10.1038\/tp.2017.136","article-title":"Dysregulation of objectively assessed 24-hour motor activity patterns as a potential marker for bipolar I disorder: results of a community-based family study.","volume":"7","author":"H Shou","year":"2017","journal-title":"Transl Psychiatry."},{"key":"ref65","article-title":"The Impact of Chronotherapeutic Interventions on Mood: A Systematic Scoping Review Across Multiple Populations.","author":"P Kahawage","journal-title":"Clin Psychol Rev."},{"key":"ref66","doi-asserted-by":"publisher","first-page":"442-448","DOI":"10.1111\/pcn.13556","article-title":"Circadian variability of objective sleep measures predicts the relapse of a mood episode in bipolar disorder: findings from the APPLE cohort.","volume":"77","author":"Y Esaki","year":"2023","journal-title":"Psychiatry Clin Neurosci."},{"key":"ref67","doi-asserted-by":"publisher","first-page":"140-149","DOI":"10.15761\/FGNAMB.1000124","article-title":"Multifractal dynamics of activity data in Bipolar Disorder: Towards automated early warning of manic relapse.","volume":"2","author":"R Heath","year":"2016","journal-title":"Fractal Geometry and Nonlinear Analysis in Medicine and Biology."},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0020650","article-title":"Scaling behavior of human locomotor activity amplitude: Association with bipolar disorder.","volume":"6","author":"P Indic","year":"2011","journal-title":"PLoS One."},{"key":"ref69","doi-asserted-by":"publisher","first-page":"2204","DOI":"10.3390\/jcm11082204","article-title":"Which Actigraphy Dimensions Predict Longitudinal Outcomes in Bipolar Disorders?","volume":"11","author":"L Ferrand","year":"2022","journal-title":"J Clin Med."},{"key":"ref107","first-page":"132-147","article-title":"Towards a better future for Canadians with bipolar disorder: principles and implementation of a community-based participatory research model.","volume":"1","author":"E Michalak","year":"2015","journal-title":"Engaged Scholar J."},{"key":"ref70","doi-asserted-by":"publisher","first-page":"1717","DOI":"10.3390\/brainsci12121717","article-title":"Validation of the Longitudinal Interval Follow-Up Evaluation for the Long-Term Measurement of Mood Symptoms in Bipolar Disorder.","volume":"12","author":"R Porter","year":"2022","journal-title":"Brain Sci."},{"key":"ref71","doi-asserted-by":"publisher","first-page":"239-244","DOI":"10.1016\/j.jad.2017.09.040","article-title":"Clinically relevant response and remission outcomes in cariprazine-treated patients with bipolar I disorder.","volume":"226","author":"W Earley","year":"2018","journal-title":"J Affect Disord."},{"key":"ref72","doi-asserted-by":"publisher","first-page":"382-389","DOI":"10.1192\/bjp.134.4.382","article-title":"A new depression scale designed to be sensitive to change.","volume":"134","author":"S Montgomery","year":"1979","journal-title":"Br J Psychiatry."},{"key":"ref73","doi-asserted-by":"publisher","first-page":"429-435","DOI":"10.1192\/bjp.133.5.429","article-title":"A rating scale for mania: reliability, validity and sensitivity.","volume":"133","author":"R Young","year":"1978","journal-title":"Br J Psychiatry."},{"key":"ref77","doi-asserted-by":"publisher","first-page":"7-117","DOI":"10.1177\/0004867420979353","article-title":"The 2020 Royal Australian and New Zealand College of Psychiatrists clinical practice guidelines for mood disorders.","volume":"55","author":"G Malhi","year":"2021","journal-title":"Aust N Z J Psychiatry."},{"key":"ref75","doi-asserted-by":"publisher","first-page":"540-548","DOI":"10.1001\/archpsyc.1987.01800180050009","article-title":"The Longitudinal Interval Follow-up Evaluation: A Comprehensive Method for Assessing Outcome in Prospective Longitudinal Studies.","volume":"44","author":"M Keller","year":"1987","journal-title":"Arch Gen Psychiatry."},{"key":"ref76","doi-asserted-by":"publisher","first-page":"402-409","DOI":"10.1016\/j.jad.2021.01.060","article-title":"Can actigraphy be used to define lithium response dimensions in bipolar disorders?","volume":"283","author":"B Etain","year":"2021","journal-title":"J Affect Disord."},{"key":"ref83","doi-asserted-by":"publisher","first-page":"944-952","DOI":"10.1037\/abn0000107","article-title":"Low social rhythm regularity predicts first onset of bipolar spectrum disorders among at-risk individuals with reward hypersensitivity.","volume":"124","author":"L Alloy","year":"2015","journal-title":"J Abnorm Psychol."},{"key":"ref84","doi-asserted-by":"publisher","first-page":"1482-1488","DOI":"10.1016\/j.jad.2021.09.037","article-title":"Validation of the youth mood recurrences risk calculator in an adult sample with bipolar disorder.","volume":"295","author":"J Fiedorowicz","year":"2021","journal-title":"J Affect Disord."},{"key":"ref85","doi-asserted-by":"publisher","first-page":"128-138","DOI":"10.1111\/bdi.12273","article-title":"Randomized, controlled trial of Interpersonal and Social Rhythm Therapy for young people with bipolar disorder.","volume":"17","author":"M Inder","year":"2015","journal-title":"Bipolar Disord."},{"key":"ref86","doi-asserted-by":"publisher","first-page":"500-508","DOI":"10.1001\/archpsyc.63.5.500","article-title":"Long-term Effectiveness and Cost of a Systematic Care Program for Bipolar Disorder.","volume":"63","author":"G Simon","year":"2006","journal-title":"Arch Gen Psychiatry."},{"key":"ref87","doi-asserted-by":"publisher","first-page":"727-740","DOI":"10.1111\/j.1399-5618.2010.00865.x","article-title":"Development of the QoL.BD: a disorder-specific scale to assess quality of life in bipolar disorder.","volume":"12","author":"E Michalak","year":"2010","journal-title":"Bipolar Disord."},{"key":"ref88","doi-asserted-by":"publisher","first-page":"33-45","DOI":"10.1016\/j.jad.2020.09.017","article-title":"The Quality of Life in Bipolar Disorder (QoL.BD) Questionnaire a decade on \u2013 A systematic review of the measurement of condition-specific aspects of quality of life in bipolar-disorder.","volume":"278","author":"E Morton","year":"2021","journal-title":"J Affect Disord."},{"key":"ref89","doi-asserted-by":"publisher","first-page":"287-302","DOI":"10.5665\/sleep.1642","article-title":"The consensus sleep diary: standardizing prospective sleep self-monitoring.","volume":"35","author":"C Carney","year":"2012","journal-title":"Sleep."},{"key":"ref90","doi-asserted-by":"publisher","first-page":"443-454","DOI":"10.1038\/nrn3743","article-title":"Light as a central modulator of circadian rhythms, sleep and affect.","volume":"15","author":"T Legates","year":"2014","journal-title":"Nature Reviews Neuroscience."},{"key":"ref92","doi-asserted-by":"publisher","first-page":"421-434","DOI":"10.1177\/1477153512450453","article-title":"Comparisons of three practical field devices used to measure personal light exposures and activity levels.","volume":"45","author":"M Figueiro","year":"2013","journal-title":"Light Res Technol."},{"key":"ref79","doi-asserted-by":"publisher","first-page":"459-464","DOI":"10.5664\/jcsm.7000","article-title":"Reliability of the Structured Clinical Interview for DSM-5 Sleep Disorders Module.","volume":"14","author":"D Taylor","year":"2018","journal-title":"J Clin Sleep Med."},{"key":"ref80","doi-asserted-by":"publisher","first-page":"e51530","DOI":"10.2196\/51530","article-title":"Evaluating the Problem of Fraudulent Participants in Health Care Research: Multimethod Pilot Study.","volume":"8","author":"V Kumarasamy","year":"2024","journal-title":"JMIR Form Res."},{"key":"ref81","doi-asserted-by":"publisher","first-page":"1266-1277","DOI":"10.1176\/appi.ajp.2011.10111704","article-title":"The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults.","volume":"168","author":"K Posner","year":"2011","journal-title":"Am J Psychiatry."},{"key":"ref82","doi-asserted-by":"publisher","first-page":"453-473","DOI":"10.1111\/j.1399-5618.2009.00726.x","article-title":"The International Society for Bipolar Disorders (ISBD) Task Force report on the nomenclature of course and outcome in bipolar disorders.","volume":"11","author":"M Tohen","year":"2009","journal-title":"Bipolar Disord."},{"key":"ref93","doi-asserted-by":"publisher","first-page":"7","DOI":"10.1186\/1740-3391-6-7","article-title":"A new approach to understanding the impact of circadian disruption on human health.","volume":"6","author":"M Rea","year":"2008","journal-title":"J Circadian Rhythms."},{"key":"ref94","doi-asserted-by":"publisher","first-page":"19110","DOI":"10.1038\/s41598-020-75622-4","article-title":"Evening home lighting adversely impacts the circadian system and sleep.","volume":"10","author":"S Cain","year":"2020","journal-title":"Sci Rep."},{"key":"ref116","doi-asserted-by":"publisher","first-page":"90-98","DOI":"10.1159\/000449417","article-title":"Meta-Analysis of the Risk of Subsequent Mood Episodes in Bipolar Disorder.","volume":"86","author":"J Radua","year":"2017","journal-title":"Psychotherapy and Psychosomatics."},{"key":"ref95","doi-asserted-by":"publisher","first-page":"27-294","DOI":"10.5705\/ss.2010.107","article-title":"Elastic net for Cox\u2019s proportional hazards model with a solution path algorithm.","volume":"22","author":"Y Wu","year":"2012","journal-title":"Stat Sin."},{"key":"ref96","doi-asserted-by":"publisher","first-page":"324-333","DOI":"10.1093\/ije\/dyu222","article-title":"Modelling recurrent events: a tutorial for analysis in epidemiology.","volume":"44","author":"L Amorim","year":"2014","journal-title":"Int J Epidemiol."},{"key":"ref97","doi-asserted-by":"publisher","first-page":"334-344","DOI":"10.1093\/ije\/dyu262","article-title":"Joint modelling of repeated measurement and time-to-event data: an introductory tutorial.","volume":"44","author":"O Asar","year":"2015","journal-title":"Int J Epidemiol."},{"key":"ref98","article-title":"Dynamic Bayesian Networks: Representation, Inference and Learning.","author":"K Murphy","year":"2002","journal-title":"Computer Science."},{"key":"ref99","doi-asserted-by":"publisher","first-page":"e56","DOI":"10.1017\/S2045796023000616","article-title":"The temporal dependencies between social, emotional and physical health factors in young people receiving mental healthcare: a dynamic Bayesian network analysis.","volume":"32","author":"F Iorfino","year":"2023","journal-title":"Epidemiol Psychiatr Sci."},{"key":"ref113","doi-asserted-by":"publisher","first-page":"109876","DOI":"10.1016\/j.yebeh.2024.109876","article-title":"User experience of a seizure risk forecasting app: A mixed methods investigation.","volume":"157","author":"R Stirling","year":"2024","journal-title":"Epilepsy Behav."},{"key":"ref100","doi-asserted-by":"crossref","first-page":"106","DOI":"10.3389\/fncom.2013.00106","article-title":"What is the mathematical description of the treated mood pattern in bipolar disorder?","volume":"7","author":"F Hadaeghi","year":"2013","journal-title":"Front Comput Neurosci."},{"key":"ref101","doi-asserted-by":"publisher","first-page":"e297","DOI":"10.1371\/journal.pmed.0040297","article-title":"Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.","volume":"4","author":"J Vandenbroucke","year":"2007","journal-title":"PLoS Med."},{"key":"ref102","doi-asserted-by":"publisher","first-page":"155-165","DOI":"10.1080\/13548506.2012.689840","article-title":"Development of an online intervention for bipolar disorder. www.moodswings.net.au.","volume":"18","author":"S Lauder","year":"2013","journal-title":"Psychol Health Med."},{"key":"ref103","first-page":"A83-A84","article-title":"www.moodswings: The highs and lows of an online intervention for bipolar disorder \u2014 preliminary findings.","volume":"42","author":"S Lauder","year":"2008","journal-title":"Aust N Z J Psychiatry."},{"key":"ref104","doi-asserted-by":"publisher","first-page":"830-844","DOI":"10.1037\/ccp0000684","article-title":"Mindfulness-based online intervention to improve quality of life in late-stage bipolar disorder: A randomized clinical trial.","volume":"89","author":"G Murray","year":"2021","journal-title":"J Consult Clin Psychol."},{"key":"ref105","doi-asserted-by":"publisher","first-page":"e32740","DOI":"10.2196\/32740","article-title":"Pilot Testing in the Wild: Feasibility, Acceptability, Usage Patterns, and Efficacy of an Integrated Web and Smartphone Platform for Bipolar II Disorder.","volume":"6","author":"K Fletcher","year":"2022","journal-title":"JMIR Form Res."},{"key":"ref106","doi-asserted-by":"publisher","first-page":"221","DOI":"10.1186\/s12888-018-1805-9","article-title":"Web-based intervention to improve quality of life in late stage bipolar disorder (ORBIT): Randomised controlled trial protocol.","volume":"18","author":"K Fletcher","year":"2018","journal-title":"BMC Psychiatry."},{"key":"ref108","doi-asserted-by":"crossref","first-page":"1-9","DOI":"10.1186\/s40345-016-0045-5","article-title":"Commentary: Harnessing the Potential of Community-Based Participatory Research Approaches in Bipolar Disorder.","volume":"4","author":"E Michalak","year":"2016","journal-title":"Int J Bipolar Disord."},{"key":"ref109","doi-asserted-by":"publisher","first-page":"77-86","DOI":"10.1016\/j.jad.2016.06.057","article-title":"Effective self-management strategies for bipolar disorder: A community-engaged Delphi Consensus Consultation study.","volume":"206","author":"E Michalak","year":"2016","journal-title":"J Affect Disord."},{"key":"ref110","doi-asserted-by":"publisher","first-page":"E1391-e1394","DOI":"10.1503\/cmaj.202066","article-title":"Problems in the deployment of machine-learned models in health care.","volume":"193","author":"J Cohen","year":"2021","journal-title":"CMAJ."},{"key":"ref111","doi-asserted-by":"publisher","first-page":"11-12","DOI":"10.1001\/jamapsychiatry.2023.4302","article-title":"Accurate machine learning prediction in psychiatry needs the right kind of information.","volume":"81","author":"B Kraus","year":"2024","journal-title":"JAMA Psychiatry."},{"key":"ref112","doi-asserted-by":"publisher","first-page":"339-347","DOI":"10.1001\/archgenpsychiatry.2010.15","article-title":"Longitudinal Course of Bipolar I Disorder: Duration of Mood Episodes.","volume":"67","author":"D Solomon","year":"2010","journal-title":"Arch Gen Psychiatry."},{"key":"ref114","doi-asserted-by":"publisher","first-page":"415-417","DOI":"10.1111\/bdi.13447","article-title":"The time has come to question the infinite maintenance treatment for bipolar disorders.","volume":"26","author":"A Dols","year":"2024","journal-title":"Bipolar Disord."},{"key":"ref115","doi-asserted-by":"publisher","first-page":"160018","DOI":"10.1038\/sdata.2016.18","article-title":"The FAIR Guiding Principles for scientific data management and stewardship.","volume":"3","author":"M Wilkinson","year":"2016","journal-title":"Sci Data."}],"container-title":["Wellcome Open Research"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2\/xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2\/iparadigms","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T10:14:09Z","timestamp":1786616049000},"score":1,"resource":{"primary":{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,13]]},"references-count":109,"URL":"https:\/\/doi.org\/10.12688\/wellcomeopenres.24189.2","relation":{"has-version":[{"id-type":"doi","id":"10.12688\/wellcomeopenres.24189.1","asserted-by":"subject"}],"has-review":[{"id-type":"doi","id":"10.21956\/wellcomeopenres.28581.r161215","asserted-by":"subject"},{"id-type":"doi","id":"10.21956\/wellcomeopenres.28581.r161215","asserted-by":"object"}]},"ISSN":["2398-502X"],"issn-type":[{"value":"2398-502X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,13]]},"assertion":[{"value":"Indexed","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2#article-reports","order":0,"name":"referee-status","label":"Referee status","group":{"name":"current-referee-status","label":"Current Referee Status"}},{"value":"10.21956\/wellcomeopenres.26682.r126221, Abigail Ortiz, University of Toronto, Toronto, Ontario, 22 Aug 2025, version 1, 1 approved, 1 approved with reservations","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v1#referee-response-126221","order":0,"name":"referee-response-126221","label":"Referee Report","group":{"name":"article-reports","label":"Article Reports"}},{"value":"<b>GREG MURRAY<\/b>; \n<i>Posted: 03 Feb 2026<\/i>; \n<b>Reviewer\u2019s comment<\/b> \n<b>Our response<\/b> \n<b>Changes to text<\/b> \n<b>REVIEWER 1<\/b>  In the abstract, what do you mean by \u201cinterepisode BD\u201d? An episodic course of illness?  \u2018Inter-episode\u2019 is a term describing people with a diagnosis of BD who do not currently meet criteria for an episode of hypomania, mania or depression [see, e.g., 1, 2]. Inter-episode is not synonymous with \u2018euthymic\u2019, which can imply an absence of residual symptoms [e.g., 3] or the presence of mood stability for a period of time [4]. &nbsp; We notice however that the more common spelling might be \u2018interepisode\u2019, so we have removed the hyphen throughout. \u2018inter-episode\u2019 replaced with \u2018interepisode\u2019 throughout  Add abbreviations to Table 1  Table 1 has been removed from Version 2 of the manuscript. We thank the reviewer for highlighting the unclear project rational in the original manuscript, and removing Table 1 improves simplicity and clarity of Version 2. &nbsp; See also response to comments #11 and #18 below. &nbsp; -  Some typos in the text \u201csome 65%...\u201d  We thank the reviewer for this feedback. A thorough proof-reading, with all authors checking Version 2. Several typographical errors have been corrected, and simplified wording applied throughout. Several editorial changes have been made, including:  \u2018Some 65%\u2019 changed to \u2018Approximately 65%\u2019 \u2018lower-middle-income\u2019 changed to \u2018lower-middle income\u2019 \u2018relative amplitude\u2019 changed to \u2018Relative Amplitude\u2019 \u2018some 60 relapses\u2019 changed to \u2018Approximately 60 relapses\u2019   Methods: why did you choose 65 years as a cut off?  We are not sure what aspect of this decision the reviewer is referring to, apologies.&nbsp; &nbsp; An upper and lower age cut-off was set to constrain between-person variance due to adolescence and later life changes in circadian and sleep function. The upper age cut-off of 65 is conventionally used to operationalise an \u2018older adult\u2019 in mental health research. Age of 65 is also a common cut-off in bipolar disorder [e.g., 5] and sleep\/circadian research [e.g., 6]. For example, Kim et al found significant differences in sleep quality with age, with decreased sleep quality amongst people aged over 65 [7, see also, 8]. Age and gender will be considered for inclusion as covariates in all analyses. &nbsp; See p.13&nbsp;  Methods: you defined euthymia using MADRS\/YMRS cutoffs but not duration. One month? Three months? Longer?  We thank the reviewer for raising this question. &nbsp; Firstly, these are cut-offs for Study 1 and Study 2 only. In Study 3, being in an acute episode is an inclusion criterion. &nbsp; We have changed the structure of the section on inclusion\/exclusion criteria to clearly separate Study 3 from Studies 1 and 2 (which largely overlap in terms of design). &nbsp; In Study 1 and 2, we do not require participants to be euthymic. Participants are required to be not experiencing an episode of bipolar disorder at baseline (because we are predicting relapse into an episode). Presence of an episode is quantified for these purposes using common cutoffs of &gt; 12 on the MADRS or &gt; 8 on the YMRS [e.g., 9, 10]. &nbsp; The reviewer is correct that some bipolar disorder studies do select participants on the basis of \n<i>mood stability<\/i>, sometimes operationalised as a low level of symptoms for an extended period\n<i> <\/i>[e.g., 4]. However, the focus of Study 1 and 2 is people with bipolar disorder who currently do not meet episode criteria. See pp. 14-15; Footnote 3  Methods: major concern is sample size to conduct the statistical analyses authors propose. Increasing the sample size is an absolute requirement to ensure all this effort results in robust findings.  We thank the reviewer for raising this concern. &nbsp;A paragraph of the Discussion is now devoted to consideration of this issue. &nbsp; The plan to address this limitation of the project is, (i) to acknowledge in advance the implications of sample sizes for the type of conclusions that can be drawn, (ii) to plan to access open data sets containing comparable prospective actigraphy (or commercial activity measure) data amongst interepisode people with bipolar disorder. To concretize the fact that sample sizes are insufficient to generate trustworthy models, we have expanded the text about our plans to cross-validate our models in data currently being collected by other groups (with whom our team has formal connections). &nbsp; The selected sample size of 100 people with BD in Study 1, 50 in Study 2 (25 BD, 25 matched controls), and 45 in Study 3 (15 manic and 30 depressed) were pragmatic maxima. We will continue to recruit beyond these numbers if funding permits. &nbsp; A power analysis on the Study 1 sample is reported to quantify our chances of demonstrating a significant effect in the Cox\u2019s survival analysis and determined that \u2018even with a 20% dropout the study is nearly sufficiently powered to detect the expected relative risk of relapse (Hazard Ratio = 1.57) and well powered to detect a slightly higher relative risk of relapse (Hazard Ratio = 1.65). &nbsp; See p.26 and pp. 38-39 &nbsp;  Methods: no mention of how authors will manage missing data, either clinical or actigraph.  We are happy to provide more detail on this. A new subsection has been added to Modelling approaches section. &nbsp; See pp. 25-26  Methods: no mention of safety precautions How will researchers manage emerging suicidality throughout the study?  We thank the reviewer for raising risk management and are happy to provide more detail. &nbsp; Note that only Study 1 and 2 are under consideration here. In Study 3, all participants are recruited from either inpatient or outpatient settings of a specialist mental health team in Christchurch, New Zealand. Study 3 participants remain under the care of this service, and existing risk management procedures will apply. &nbsp; Key principles of the approach include protection of autonomy where possible, explicit assignment of risk management to participants and their local networks, and advance planning for crises [11]. Text edited and expanded in \n<i>Risk Management Approach. <\/i>The new text more clearly distinguishes between management processes for \n<i>acute suicidality<\/i> versus management of a participant in whom suicidality or other clinical risk issues emerge through their involvement. &nbsp; See pp.31-34  Methods: If those in Study 1 and 2 are going to be euthymic, there is no need to exclude those with psychosis  Agreed. By definition, a person cannot meet eligibility criteria if they are psychotic outside an episode or are currently in an episode. Reference to acute psychosis exclusion removed.  Methods: why are sleep or movement d\/o not an exclusion criterion for Study 1 and 2?  The reviewer identifies two issues that were not adequately explained in the text, (i) the role of assessment of sleep disorders and other comorbidities, and (ii) exclusion criteria based on potential confounding of the actigraphy signal. &nbsp; Having a physical problem that might confound RAR interpretation is an exclusion in all three studies and is a common exclusion in actigraphy research. The confusion in our text was that \n<i>different examples of physical problems that might impact RAR interpretation were provided for Study 1 and 2 compared with Study 3<\/i>. Text has been revised and some new text added. &nbsp; Presence of common BD comorbidities is not an exclusion criterion. Anxiety disorders, substance use disorders and sleep disorders will be assessed by structured clinical interview at baseline and explored for potential inclusion in predictive models. This approach allows us to build models that generalise to complex clinical presentations while still accounting for their impact on models. &nbsp; See last para p.13; first para p.14 11. Methods: why is \u201ccurrent intoxication or withdrawal from illicit substances&quot; not an exclusion criteria for Study 1 and 2? We thank the reviewer for picking up this ambiguity. The reviewer has highlighted that the overall logic of the multi-study design can be made clearer. &nbsp; The exclusion for &quot;current intoxication or withdrawal&quot; is only relevant in&nbsp;Study 3, in which participants are primarily recruited from an&nbsp;inpatient setting. It is not uncommon for individuals to be admitted to hospital in a state of intoxication or acute withdrawal (e.g., from alcohol or methamphetamines). These states would profoundly impact the capacity to consent and confound the actigraphy data collected in the initial days. Therefore, as a standard procedural step in this inpatient setting, potential participants are only approached for recruitment&nbsp;after&nbsp;this acute phase (~72 hours) has passed. Text signposting the project- versus study-level differences has been added, as has a new Box which compares and contrasts the three studies\u2019 methods at a glance. &nbsp; See first para. p15; Box 1 on p.45 &nbsp; &nbsp; &nbsp;  Methods: See above (#10) \u2013 does that mean that comorbid substance use or abuse is allowed for the whole sample?  We thank the reviewer for this request for clarification about the measurement of comorbid diagnoses. See response to point 11 above. Additional text added to Inclusion and Exclusion criteria Studies 1 and 2, new Box added. &nbsp; The two most prevalent psychiatric comorbidities of BD (anxiety and substance disorders), and comorbid sleep disorders are rigorously assessed by clinical interview in all three studies. Comorbid substance use\/abuse within the last 12 months (measured on the SCID for those who screen as potentially impacted by self-report AUDIT and DUDIT), anxiety disorders (measured on the SCID) and sleep disorders (via SCISD-R). To maximise generalisability of the findings, presence of these disorders is not an exclusion criterion. &nbsp; The reviewer is correct that these comorbidities can impact sleep and circadian function. Anxiety, substance and sleep disorders will be investigated as potential covariates in analyses. See first para, p.14 &nbsp; &nbsp; &nbsp; &nbsp;  &nbsp;Methods: Authors talk repeatedly about \u201cseveral ML\u201d, but only mention one.  &nbsp; 13.1 This is particularly important for Study 2, Research Question 2: \u201cDo any ML models enable\u2026\u201d. We thank the reviewer for this observation. The text has been expanded to clarify that multiple complementary machine learning (ML) methods will be employed, each addressing a distinct analytical aim. &nbsp; &nbsp; &nbsp; See p.25, pp.27-28   &nbsp; 13.2 Clarify which variables will be your input (e.g., demographic, clinical, actigraph) and what you will be your output (predicting an episode) We agree that this could have been made clearer in the original MS.&nbsp; In Version 2, several changes have been made, including a new subsection headed \u2018\n<b>Input and outcome variables\u2019<\/b> &nbsp; &nbsp; See pp.30-31; p.24-25  Methods: Clarify why 26 weeks is your predictive horizon. Why not shorter\/larger windows?  We thank the reviewer for requesting this clarification about Study 1 and 2 timeframes. &nbsp; The dynamics of bipolar disorder course have been investigated across timescales from days [emotion regulation, e.g., 12] to weeks [mood instability, rapid cycling, 13, 14] to years [predominant polarity, seasonality, lifetime course, 15]. &nbsp;Research aiming to predict relapse events has tended to use longer predictive horizons [e.g., 3 years, 16]. The total predictive horizon for participants in Study 1 here is 78 weeks, assessed in six 13-week assessment windows (i.e., 3 X 26 week epochs, with two assessment interviews per epoch). &nbsp; As mentioned in the first paragraph of the Introduction, the focus here is on \u201cimproving forecasting of the short-term course\u2019 (weeks and months)\u201d. This a clinically important timeframe:&nbsp; the algorithm identified here could form part of a risk calculator for clinicians and patients. &nbsp; Text clarified \u2013 brief rationale for 26-week predictive horizon added. &nbsp;&nbsp;See p.16  Methods: What about episodes that occur between your interviews?  We thank the reviewer for highlighting a central design feature that we did not make clear. New text has been added in Version 2 (see also #5 above), along with further explication of the LIFE. &nbsp;  The primary outcome in Study 1 and 2 is measured on the \n<i>retrospective<\/i> LIFE interview conducted at 13 and 26 weeks per epoch. The primary purpose of the LIFE interview is to provide a medium-term measure of mood states across time, avoiding the participant burden of \n<i>prospective self-monitoring <\/i>of symptoms, and the problem that increasing symptoms are associated with growing impairment of insight. As described in the section \n<b><i>Primary outcome measures, <\/i><\/b>the LIFE generates a weekly mood score, providing an estimate of presence versus absence of an episode per week in the retrospective window. The reviewer highlights that there could be some confusion around the fact that Study 1 and Study 2 employ two very different measures of presence of an episode (MADRS\/YMRS cut-offs versus LIFE retrospective estimates of weekly episode status). We have added a footnote to address this ambiguity.  See pp.22-23, including Footnote 5; see Footnote 4 on p.14.  Methods: Include more information about Hadaeghi\u2019s complex system model of relapse  We thank the reviewer for these insightful comments (15, 16,17). The Complex systems analysis subsection has been rewritten to improve accessibility for a clinical readership. It has also been expanded to clarify the model inputs and nonlinear features to be examined. &nbsp; We thank the reviewer for this suggestion.&nbsp; New text has been added to explain Hadaeghi\u2019s Complex System Model. This model conceptualises bipolar disorder (BD) as a dynamic system with two interacting subsystems: the circadian pacemaker (sleep\u2013wake regulation) and the frontal\u2013limbic emotional regulation network. Disturbances in either subsystem, or in their feedback loops, can push the system toward a critical transition or \u201ctipping point,\u201d corresponding clinically to a relapse into mania or depression. The new material includes explicit statement about the types of nonlinear features that are likely to appear in analyses.&nbsp; &nbsp; &nbsp; See pp.28-30  Methods: See above (#16): Your audience will be clinical \u2013 explain what you mean by \u201ctrained on statistical, time-frequency, and nonlinear features\u201d.   Methods: Clarify which nonlinear features are you planning to use. There are many.   Methods: Treatment response comes as a surprise in Study 3.&nbsp;  &nbsp; &nbsp; We thank the reviewer for requesting more detail on this issue. The original manuscript does include some words about the role of Study 3, but this comment has encouraged us to conduct a root and branch review of this aspect of the project. &nbsp;&nbsp;&nbsp; &nbsp; From a complex systems viewpoint, relapse into a new episode is a critical state transition \u2013 metaphorically a tipping point. Our primary clinical objective in is to develop a predictive algorithm for this transition. To ensure this algorithm is robust and mechanistically informed, however, our project design includes a complementary investigation of an alternative state transition, namely, recovery from an acute episode. &nbsp; &nbsp; \n<i>Studying this second transition is not the primary clinical goal but is scientifically important. It allows us to test whether the signals predicting relapse are markers of state transition in general, thereby increasing confidence in their validity. It also provides a unique window into the system's dynamics during a period of rapid change, offering mechanistic insights that may refine the primary relapse model.<\/i> New text has been added at multiple points, and Abstract rewritten to elevate the notion of state transition in the project logic. Text introducing the project at the end of the Introduction has been revised to highlight the mechanistic significance of investigating whether dynamic models predicting relapse (from Study 1) extend to prediction of recovery (Study 3). &nbsp; See Abstract; p.7; pp.11-12; pp.18-19; p.38; revised Figure 3. &nbsp; &nbsp; &nbsp; &nbsp; Reviewer comment 18 continued:  Please enumerate scales you plan to use to measure treatment response; How will you account for polypharmacy in the model; How will you ensure treatment compliance?  We agree that the description of measurement of response to treatment in Study 3 was ambiguous in text and Figure 3. &nbsp;Figure 3 has now been redrawn to focus only on the primary aim of Study 3.&nbsp; &nbsp;&nbsp; &nbsp; Study 3 has primary, secondary, and tertiary aims. The primary aim is to investigate the extension of the algorithm informed by Hadaeghi\u2019s Complex System Model of relapse prediction (trained and tested on Study 1 data) to response to treatment, with both relapse (Study 1 and 2 outcome) and treatment response (Study 3 outcome) being understood as state transitions (Hadaeghi et al., 2016). &nbsp; The state transition of Study 3 is measured in a binary variable, namely, response to treatment operationalised as 50% reduction in MADRS (for those in depressive episode at baseline) or 50% reduction in YMRS (for those in manic episode at baseline) at 8 weeks. Participants will wear actigraphs for 21 days \u2013 for consistency with Study 1 and 2, the first 14 days only will be used in modelling of binary response to treatment. &nbsp; Study 3 is a naturalistic study of treatment in the real world, and important covariates can be measured but not controlled.&nbsp; Polypharmacy will be drawn from hospital records. Treatment compliance cannot be ensured, but we will record any notes about treatment compliance from hospital records for consideration.&nbsp;&nbsp; &nbsp; &nbsp;See updated p.18 and Figure 3 19.1 This part is very confusing \u2013 I assume it is an output but there is a gap: what will you be measuring that will permit treatment response to be understood as a state transition? Agree. See new text added in response to #13.2 above  Methods: Secondary Aim of Study 3 is unclear. Please clarify.  Agree. Wording of secondary aim clarified, with primary, secondary and tertiary aims of Study 3 now distinguished. Text corrected throughout and Figure 3 redrawn to accurate reflect Study 3\u2019s primary aim, and the primary aim only.&nbsp; &nbsp; &nbsp; &nbsp; See pp.18-19; new Figure 3.  More detail about the integrated algorithm for risk relapse is needed  We are happy to provide more detail. \n<b>Work Package 4: Integrative Data Synthesis<\/b> section has been rewritten and expanded.&nbsp; &nbsp; It is important to note that it is not possible to predict in advance what specific variables will warrant inclusion in models \u2013 this will depend on the distributions of collected variables and their interrelationships [20]. The project is therefore exploratory. &nbsp;The Editor of Wellcome Open has agreed that we will submit a detailed data analysis plan only after data has been collected, screened and cleaned. See p.20-21  Fig 3 is different from what is written: will assessments be done weekly, as per Figure; or at week 8, as per text?  We apologise for this ambiguity, which has been corrected. Figure 3 has been redrawn, and text corrected.&nbsp; See response to #19 above.  Why are you only considering residual depressive symptoms as predictors and not residual anxiety or residual cognitive impairment \u2013 as these are also common clinical characteristics that indicate a potentially better treatment response to anticonvulsants or antipsychotics?  We thank the reviewer for picking up on this ambiguity across studies.&nbsp; The clinical predictors we are modelling are described for Study 1 and 2 only, and Version 2 makes this clear.&nbsp; Text has been clarified to ensure the reader knows which studies include the five clinical predictors. &nbsp; We are not planning to add clinical predictors into the models of Study 3 given the small sample size, and concerns about participant burden. Note that, as the reviewer comments, candidate predictors of response to treatment are not the same as candidate predictors of relapse. &nbsp; See our response to #23.1 below for further discussion of our choice of five specific \u2018clinical predictors\u2019 to include in models of Study 1 and 2. See pp.24-25 23.1 Please justify your choice of these 5 clinical predictors These five clinical predictors are only being used in Studies 1 and 2, not the Study 3 prediction of treatment response (these are evidence-based predictors of relapse, not response to treatment). The five chosen predictors have now been explicitly justified in new text. &nbsp; Given the focus here on modelling analyses, we cannot predict in advance what specific variables will warrant inclusion in models [20]. The project is exploratory, and the \n<i>Editor of Wellcome Open has accepted our suggestion that we will submit a detailed data analysis plan only after we have collected, screened and cleaned data<\/i>. See p.24-25  Is the rationale for, and objectives of, the study clearly described? YES  - -  Is the study design appropriate for the research question? YES  - -  Are sufficient details of the methods provided to allow replication by others? NO  We agree that the manuscript provides insufficient details for replications. We had mentioned in Footnote 1 that methodological details of each study would be provided in separate study-level protocol papers, but we agree with the reviewer that this manuscript needs to be more explicit about the relationship between project-level and study-level decisions. &nbsp; Footnote 1 has been removed, and the explanation for the project-level protocol paper moved to text and expanded in an intro paragraph to the Protocol section &nbsp; See p.12 and Box 1 &nbsp; \n<b>References<\/b> 1.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Gershon, A., et al., \n<i>Daily Actigraphy Profiles Distinguish Depressive and Interepisode States in Bipolar Disorder.<\/i> Clin Psychol Sci, 2016. \n<b>4<\/b>(4): p. 641-650. 2.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ng, T.H., et al., \n<i>Sleep\u2013wake disturbance in interepisode bipolar disorder and high-risk individuals: A systematic review and meta-analysis.<\/i> Sleep Medicine Reviews, 2015. \n<b>20<\/b>: p. 46-58. 3.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fujita, A., et al., \n<i>Comparison of objective and subjective sleep parameters in patients with bipolar disorder in both euthymic and residual symptomatic periods.<\/i> J Psychiatr Res, 2021. \n<b>145<\/b>: p. 190-196. 4.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Yeom, J.W., et al., \n<i>Daily Self-Monitoring and Feedback of Circadian Rhythm Measures in Major Depression and Bipolar Disorder Using Wearable Devices and Smartphones-The Circadian Rhythm for Mood (CRM&reg;) Trial Protocol: A Randomized Sham Controlled Double-Blind Trial.<\/i> Psychiatry Investig, 2024. \n<b>21<\/b>(8): p. 918-924. 5.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Murray, G., et al., \n<i>Online mindfulness-based intervention for late-stage bipolar disorder: pilot evidence for feasibility and effectiveness.<\/i> J Affect Disord, 2015. \n<b>178<\/b>: p. 46-51. 6.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kahawage, P., et al., \n<i>Social Rhythm Disruption is Associated with Greater Depressive Symptoms in People with Mood Disorders: Findings from a Multinational Online Survey During COVID-19.<\/i> Can J Psychiatry, 2022. \n<b>67<\/b>(11): p. 831-840. 7.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kim, M., et al., \n<i>Association Between Age and Sleep Quality: Findings From a Community Health Survey.<\/i> Sleep Med Res, 2021. \n<b>12<\/b>(2): p. 155-160. 8.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hood, S. and S. Amir, \n<i>The aging clock: circadian rhythms and later life.<\/i> J Clin Invest, 2017. \n<b>127<\/b>(2): p. 437-446. 9.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Earley, W., et al., \n<i>Clinically relevant response and remission outcomes in cariprazine-treated patients with bipolar I disorder.<\/i> J Affect Disord, 2018. \n<b>226<\/b>: p. 239-244. 10.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Porter, R.J., et al., \n<i>Validation of the Longitudinal Interval Follow-Up Evaluation for the Long-Term Measurement of Mood Symptoms in Bipolar Disorder.<\/i> Brain Sciences, 2022. \n<b>12<\/b>(12): p. 1717. 11.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fletcher, K., et al., \n<i>Web-based intervention to improve quality of life in late stage bipolar disorder (ORBIT): Randomised controlled trial protocol.<\/i> BMC Psychiatry, 2018. \n<b>18<\/b>(1): p. 221. 12.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kim, J.A., et al., \n<i>Chronotherapeutic intervention targeting emotion regulation brain circuitry, symptoms, and suicide risk in adolescents and young adults with bipolar disorder: a pilot randomised trial.<\/i> BMJ Ment Health, 2025. \n<b>28<\/b>(1). 13.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Stromberg, A.R., et al., \n<i>Modeling and predicting mood instability in a longitudinal cohort of bipolar disorder.<\/i> Nature Mental Health, 2025. \n<b>3<\/b>(10): p. 1267-1275. 14.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Schneck, C.D., et al., \n<i>The prospective course of rapid-cycling bipolar disorder: findings from the STEP-BD.<\/i> Am J Psychiatry, 2008. \n<b>165<\/b>(3): p. 370-7; quiz 410. 15.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fiedorowicz, J.G., et al., \n<i>Validation of the youth mood recurrences risk calculator in an adult sample with bipolar disorder.<\/i> J Affect Disord, 2021. \n<b>295<\/b>: p. 1482-1488. 16.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Etain, B., et al., \n<i>Clinical predictors of recurrences in bipolar disorders type 1 and 2: A FACE-BD longitudinal study.<\/i> J Psychiatr Res, 2021. \n<b>134<\/b>: p. 129-137. 17.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Uher, R., et al., \n<i>Clinical course predicts long-term outcomes in bipolar disorder.<\/i> Psychological Medicine, 2019. \n<b>49<\/b>(7): p. 1109-1117. 18.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Radua, J., H. Grunze, and B.L. Amann, \n<i>Meta-Analysis of the Risk of Subsequent Mood Episodes in Bipolar Disorder.<\/i> Psychotherapy and Psychosomatics, 2017. \n<b>86<\/b>(2): p. 90-98. 19.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ferrand, L., et al., \n<i>Which Actigraphy Dimensions Predict Longitudinal Outcomes in Bipolar Disorders?<\/i> J Clin Med, 2022. \n<b>11<\/b>(8). 20.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ioannidis, J.P.A., \n<i>Pre-registration of mathematical models.<\/i> Mathematical Biosciences, 2022. \n<b>345<\/b>: p. 108782.","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2#referee-comment-6415","order":1,"name":"referee-comment-6415","label":"Referee Comment","group":{"name":"article-reports","label":"Article Reports"}},{"value":"10.21956\/wellcomeopenres.26682.r126730, Danielle Novick, Pittsburgh Veterans Affairs Medical Center, Pittsburgh, 23 Aug 2025, version 1, 1 approved, 1 approved with reservations","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v1#referee-response-126730","order":2,"name":"referee-response-126730","label":"Referee Report","group":{"name":"article-reports","label":"Article Reports"}},{"value":"<b>GREG MURRAY<\/b>; \n<i>Posted: 03 Feb 2026<\/i>; REVIEWER 2 In this manuscript, the authors detail a new project that aims to collect rest-activity rhythms (RAR) by actigraphy from individuals with bipolar disorder (BD) across two sites, one in Australia and one in India.&nbsp; The project also includes four dissertation projects. Note that there is also a study located in New Zealand (Study 3). We have made some additions to the text to make the rationale and aims of Study 3 clearer (please see responses to Reviewer 1). The goals are two-fold: to develop an algorithm of relapse risk based on RAR data and to further understand RAR data in BD. Correct.  - Overall, the manuscript is well written, clear, and concise.&nbsp; The abstract is appropriate in content; there is a small typo (see below). The introduction clearly introduces relevant ideas and what is to follow.&nbsp; The authors\u2019 explanations of topics central to understanding the protocol is appropriate.&nbsp; The methodology seems reasonable and sound, and statistical analyses appropriate (although, admittedly, outside the expertise of this reviewer). Thank you.  - Typo in Abstract: Three prospective studies will enrol adults living with BD I or II. Study 1 (Australia) will recruit a sample of N = 100 Not sure what the typo is here?&nbsp; We have italicised the \u2018N\u2019 because it is standard practice to italicise statistical terms.  - Is the rationale for, and objectives of, the study clearly described? YES - - Is the study design appropriate for the research question? YES - - Are sufficient details of the methods provided to allow replication by others? YES - - Are the datasets clearly presented in a useable and accessible format? YES - - &nbsp; &nbsp; &nbsp; &nbsp;  \n<b>References<\/b> 1.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Gershon, A., et al., \n<i>Daily Actigraphy Profiles Distinguish Depressive and Interepisode States in Bipolar Disorder.<\/i> Clin Psychol Sci, 2016. \n<b>4<\/b>(4): p. 641-650. 2.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ng, T.H., et al., \n<i>Sleep\u2013wake disturbance in interepisode bipolar disorder and high-risk individuals: A systematic review and meta-analysis.<\/i> Sleep Medicine Reviews, 2015. \n<b>20<\/b>: p. 46-58. 3.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fujita, A., et al., \n<i>Comparison of objective and subjective sleep parameters in patients with bipolar disorder in both euthymic and residual symptomatic periods.<\/i> J Psychiatr Res, 2021. \n<b>145<\/b>: p. 190-196. 4.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Yeom, J.W., et al., \n<i>Daily Self-Monitoring and Feedback of Circadian Rhythm Measures in Major Depression and Bipolar Disorder Using Wearable Devices and Smartphones-The Circadian Rhythm for Mood (CRM&reg;) Trial Protocol: A Randomized Sham Controlled Double-Blind Trial.<\/i> Psychiatry Investig, 2024. \n<b>21<\/b>(8): p. 918-924. 5.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Murray, G., et al., \n<i>Online mindfulness-based intervention for late-stage bipolar disorder: pilot evidence for feasibility and effectiveness.<\/i> J Affect Disord, 2015. \n<b>178<\/b>: p. 46-51. 6.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kahawage, P., et al., \n<i>Social Rhythm Disruption is Associated with Greater Depressive Symptoms in People with Mood Disorders: Findings from a Multinational Online Survey During COVID-19.<\/i> Can J Psychiatry, 2022. \n<b>67<\/b>(11): p. 831-840. 7.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kim, M., et al., \n<i>Association Between Age and Sleep Quality: Findings From a Community Health Survey.<\/i> Sleep Med Res, 2021. \n<b>12<\/b>(2): p. 155-160. 8.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hood, S. and S. Amir, \n<i>The aging clock: circadian rhythms and later life.<\/i> J Clin Invest, 2017. \n<b>127<\/b>(2): p. 437-446. 9.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Earley, W., et al., \n<i>Clinically relevant response and remission outcomes in cariprazine-treated patients with bipolar I disorder.<\/i> J Affect Disord, 2018. \n<b>226<\/b>: p. 239-244. 10.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Porter, R.J., et al., \n<i>Validation of the Longitudinal Interval Follow-Up Evaluation for the Long-Term Measurement of Mood Symptoms in Bipolar Disorder.<\/i> Brain Sciences, 2022. \n<b>12<\/b>(12): p. 1717. 11.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fletcher, K., et al., \n<i>Web-based intervention to improve quality of life in late stage bipolar disorder (ORBIT): Randomised controlled trial protocol.<\/i> BMC Psychiatry, 2018. \n<b>18<\/b>(1): p. 221. 12.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kim, J.A., et al., \n<i>Chronotherapeutic intervention targeting emotion regulation brain circuitry, symptoms, and suicide risk in adolescents and young adults with bipolar disorder: a pilot randomised trial.<\/i> BMJ Ment Health, 2025. \n<b>28<\/b>(1). 13.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Stromberg, A.R., et al., \n<i>Modeling and predicting mood instability in a longitudinal cohort of bipolar disorder.<\/i> Nature Mental Health, 2025. \n<b>3<\/b>(10): p. 1267-1275. 14.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Schneck, C.D., et al., \n<i>The prospective course of rapid-cycling bipolar disorder: findings from the STEP-BD.<\/i> Am J Psychiatry, 2008. \n<b>165<\/b>(3): p. 370-7; quiz 410. 15.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fiedorowicz, J.G., et al., \n<i>Validation of the youth mood recurrences risk calculator in an adult sample with bipolar disorder.<\/i> J Affect Disord, 2021. \n<b>295<\/b>: p. 1482-1488. 16.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Etain, B., et al., \n<i>Clinical predictors of recurrences in bipolar disorders type 1 and 2: A FACE-BD longitudinal study.<\/i> J Psychiatr Res, 2021. \n<b>134<\/b>: p. 129-137. 17.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Uher, R., et al., \n<i>Clinical course predicts long-term outcomes in bipolar disorder.<\/i> Psychological Medicine, 2019. \n<b>49<\/b>(7): p. 1109-1117. 18.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Radua, J., H. Grunze, and B.L. Amann, \n<i>Meta-Analysis of the Risk of Subsequent Mood Episodes in Bipolar Disorder.<\/i> Psychotherapy and Psychosomatics, 2017. \n<b>86<\/b>(2): p. 90-98. 19.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ferrand, L., et al., \n<i>Which Actigraphy Dimensions Predict Longitudinal Outcomes in Bipolar Disorders?<\/i> J Clin Med, 2022. \n<b>11<\/b>(8). 20.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Ioannidis, J.P.A., \n<i>Pre-registration of mathematical models.<\/i> Mathematical Biosciences, 2022. \n<b>345<\/b>: p. 108782.","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2#referee-comment-6416","order":3,"name":"referee-comment-6416","label":"Referee Comment","group":{"name":"article-reports","label":"Article Reports"}},{"value":"10.21956\/wellcomeopenres.28581.r161215, Yeliz Akta\u015f, \u0130zmir Atat\u00fcrk Training and Research Hospital, \u0130zmir, 31 Jul 2026, version 2, indexed","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2#referee-response-161215","order":4,"name":"referee-response-161215","label":"Referee Report","group":{"name":"article-reports","label":"Article Reports"}},{"value":"<b>GREG MURRAY<\/b>; \n<i>Posted: 12 Aug 2026<\/i>; Authors' responses to Reviewer 3:&nbsp; &nbsp;&nbsp;  I would like to congratulate the authors on this well-conceived and innovative study protocol. The integration of actigraphy, circadian rhythm assessment, and machine learning to predict relapse in bipolar disorder addresses an important and clinically relevant research question. The multicenter, international design further strengthens the potential impact and generalizability of the study. Overall, the protocol is well presented, and I hope the following comments will help further improve the manuscript.  \n<b>Response:<\/b>&nbsp;We thank the reviewer for this encouraging feedback.&nbsp; Wellcome Open has informed us that Version 2 of the protocol paper has now passed peer review and been published.&nbsp; Nonetheless, we greatly appreciate Reviewer 3\u2019s insightful comments and provide detailed responses below.&nbsp;&nbsp;  1.&nbsp;Participants will continue receiving treatment as usual throughout follow-up, and treatment modifications are likely to occur in response to emerging prodromal symptoms. Such changes (e.g., medication adjustments, initiation of psychotherapy, hospitalization, or other clinical interventions) may alter both rest-activity rhythms and relapse risk, introducing time-varying confounding. The protocol should clarify how treatment changes during follow-up will be recorded and whether they will be incorporated into the predictive models or addressed using appropriate statistical methods for time-dependent confounding.  \n<b>Response:&nbsp;<\/b>We thank the reviewer for this important observation.&nbsp; In Studies 1 and 2, comprehensive assessment of current medications is conducted at each of the three burst periods (Baseline, Follow-up 2 and Follow-up 4 in Australia; Baseline in India). In Study 3, medication prescribing will be recorded once weekly from the hospital medication chart. &nbsp;If warranted, variables from these assessments may be explored for inclusion in predictive models as the reviewer suggests.&nbsp;  2.&nbsp;The proposed sample size appears limited, particularly for Study 2 (25 patients with bipolar disorder) and Study 3 (45 patients), considering the complexity of the planned machine learning approaches. Although this is a study protocol, the authors should provide a stronger justification for the adequacy of these sample sizes for model development and external validation. In addition, it would be helpful to clarify the strategies that will be used to minimize overfitting and ensure model robustness, particularly given the relatively small datasets available for these analyses.  \n<b>Response:&nbsp;<\/b>We thank the reviewer for these important observations. The manuscript notes that the most important limitation of the project is the relatively small sample sizes for computational modelling. Several strategies are embedded to mitigate the sample size constraint and maximise the project\u2019s value. First, it must be recognised that the high-resolution nature of actigraphy (&gt; one million data points per participant per epoch in Study 1) provides a large amount of within-subject data for modelling within-subject dynamics. Relatedly, Study 1 collects three distinct monitoring epochs per participant, increasing the number of analytic events available for intra-individual analyses. Indeed, with the planned 14 X 3 X 100 = 4200 person days across 18 months participation, Study 1 is one of the largest prospective actigraphy studies in bipolar disorder.&nbsp; However, within-subject data density does not completely make up for limited sample size.&nbsp;&nbsp; Second, we acknowledge that confidence in the models developed here must await replication of the project\u2019s findings in unseen data sets. We are in discussion with teams worldwide for data sharing to support these tests. For example, we have had discussions with, (i) another Wellcome-funded project collecting RAR data via actigraphy in a large prospective sample of BD https:\/\/wellcome.org\/research-funding\/funding-portfolio\/funded-grants\/ambient-and-passive-collection-sleep-and-circadian (ii) the BD2 consortium which is collecting RAR via Fitbit in a large prospective study of people diagnosed with BD-I https:\/\/www.bipolardiscoveries.org\/our-work\/integrated-network\/. Prompted by Reviewer 3\u2019s concern, we have now conducted power analyses for Study 2 and 3. For Study 2 a power analysis for an ANCOVA analysis predicting the RAR signal developed in Study 1 (e.g. Relative Amplitude) using a model including a group variable (BD v healthy control) and the chosen continuous clock gene variable (e.g. maximum amplitude of diurnal rhythm) was considered using G-Power 3.1.9.7. For a sample size of 50 with 25 people in each group, we can detect strong associations between the RAR and clock gene variables with 80% power and 5% significance (Cohen\u2019s f=0.4045, r=0.375 for the entire sample (n=50) and r=0.456 for the BD sample (n=25)). Takahashi et al. [1] found a positive correlation between accelerometer-measured physical activity and amplitude of PER3 expression in hair follicle cells among healthy older adults (r = 0.58, n = 20). This is a different tissue and a simpler activity measure than our RAR variable, but the result suggests that for the above sample size there is sufficient power to detect an association between the activity-clock gene and the optimised RAR measure developed in Study 1. For Study 3 a binary logistic regression analysis predicting a positive response to treatment from the RAR variable developed in Study 1 was considered for the power analysis. It has previously been found that adding objective 24-hour rest-activity circadian rhythms metrics (actigraphy) to clinical baselines (e.g. age, sex, BD subtype, number of mood stabilizers) improves prediction Areas Under the Curve (pAUCs) significantly (e.g. from 0.64 to 0.82)\n<sup> <\/sup>[2]. Using MedCalc Software [3] and assuming sample sizes of 30 and 15 for depressed and manic people, with expected positive response rates of 40% and 50% [4], a power analysis was conducted. The results suggest that, for the above sample sizes, a slightly larger AUC improvement (from 0.64 to 0.85) can be detected with 80% power at a 5% significance level using our optimised RAR measure. This suggests that the sample size for Study 3 is also sufficient. To minimize the risk of overfitting, we will employ model-agnostic validation procedures and model-specific regularization strategies. Model development and evaluation will use nested cross-validation, with data partitioned at the participant level such that all observations from a given participant remain within the same fold. The inner cross-validation loop will be used for model and hyperparameter selection, while performance will be estimated exclusively from predictions for participants held out in the outer loop. All preprocessing and model-selection steps will be performed within the corresponding training folds to avoid information leakage. In addition, model complexity will be deliberately constrained in keeping with our emphasis on interpretability. Regularization techniques, including dropout where applicable, together with early stopping will be used to further limit overfitting and reliance on unstable or overly complex feature combinations. Finally, the manuscript explains that \u201c\u2026the selected sample sizes of N = 100 people with BD in Study 1, 50 in Study 2 (25 BD, 25 matched controls), and 45 in Study 3 (15 manic and 30 depressed) were \n<b>pragmatic maxima.<\/b> Recruitment beyond these pragmatic maxima will continue if resourcing permits.\u201d In the funding application, we did not want to over-promise on recruitment and wanted reviewers to have a realistic sense of what could be achieved with our existing recruitment sources. &nbsp;We are happy to report that recruitment in all three studies has met or exceeded our expected targets to date, so we are likely to exceed these targets to some degree. Nonetheless, as the project\u2019s major limitation (recognised by Reviewers 2 and 3), we would like to further increase sample size in all three studies by extending the project timeline (to be discussed with the funder Wellcome Trust UK).  3. The power analysis assumes a 20% attrition rate; however, the protocol does not clearly describe how participant dropout and incomplete follow-up will be managed during the 78-week follow-up period. Given the lengthy follow-up and repeated actigraphy assessments, participant attrition is likely. The authors should clarify how withdrawals, loss to follow-up, and incomplete actigraphy data will be handled in the primary analyses, as well as describe any planned retention strategies to minimize missing data.  \n<b>Response:&nbsp;<\/b>The Study 1 Patient Information and Consent Form describes that participants are free to withdraw at any time. &nbsp;\u201cYou are free to stop taking part in the study at any stage. If you withdraw, data you have already contributed to the study cannot be withdrawn... If you did decide to withdraw, we would like to invite you to an optional interview to help us understand people\u2019s experience of the study.\u201d The approach to incomplete device data (actigraphy or light sensor) is described in the Data Preprocessing section of the manuscript (p.14). A range of retention strategies have been adopted for the Australian Study 1.&nbsp; First, participants are appropriately reimbursed not just for the time commitment of active involvement (e.g., interviews, completing online surveys), but also for the inconvenience associated with wearing actigraph and light sensor during Burst Monitoring periods. Participants who complete all components of the study receive $1202.50 reimbursement, plus entry into a 1 in 10 chance prize draw for a further $1000.00. Second, participant burden has been aggressively minimised throughout by including only essential assessments for achieving project aims. Third, each participant is supported through the entire 18-month study by their own Research Assistant, who is also contactable by phone during the three Burst Monitoring periods. Fourth, participants can be \u2018paused\u2019 in the study on the basis of ill-health (e.g., relapse event), with the option to be recontacted in four weeks to see if they are willing and able to recommence. &nbsp;Finally, we have an explicit operating principle that participation will be a net benefit: should there be a tension between a participant\u2019s wellbeing and the quantity of data collected, the former will always be elevated. &nbsp;&nbsp;&nbsp; As a set, each of these strategies falls under the umbrella of empowered participation (see also our management of clinical risk), the principles of which have been developed in close consultation with our lived experience experts. As shown in the language and priorities described in recruitment materials, we see participants as lived experience collaborators, joining us on a project to improve outcomes for people with bipolar disorder.&nbsp;  [1] Takahashi M, Haraguchi A, Tahara Y, Aoki N, Fukazawa M, Tanisawa K, et al. Positive association between physical activity and PER3 expression in older adults. Sci Rep. 2017;7(1):39771. [2] Ferrand L, Hennion V, Godin O,&nbsp;\n<i>et al.<\/i>: Which Actigraphy Dimensions Predict Longitudinal Outcomes in Bipolar Disorders?&nbsp;\n<i>J Clin Med.<\/i>&nbsp;2022;&nbsp;\n<b>11<\/b>(8): 2204. [3] MedCalc Software Ltd. Power calculator for Area under ROC curve. https:\/\/www.medcalc.org\/en\/calc\/power-area-under-roc-curve.php (Version 23.6.5; accessed August 10, 2026) [4] Averill IR, Crowe M, Frampton CM, Beaglehole B, Lacey CJ, Jordan J, Wilson LD, Douglas KM, Porter RJ. Clinical response to treatment in inpatients with depression correlates with changes in activity levels and psychomotor speed. Aust N Z J","URL":"https:\/\/wellcomeopenresearch.org\/articles\/10-320\/v2#referee-comment-6697","order":5,"name":"referee-comment-6697","label":"Referee Comment","group":{"name":"article-reports","label":"Article Reports"}},{"value":"This work was supported by Wellcome [226945]. The funder provided suggestions for data harmonization, but the authorship team was granted full independence in making decisions about the design and procedures for this trial. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.","order":6,"name":"grant-information","label":"Grant Information"},{"value":"This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.","order":0,"name":"copyright-info","label":"Copyright"}]}}