{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T18:15:16Z","timestamp":1783707316562,"version":"3.55.0"},"reference-count":52,"publisher":"Public Library of Science (PLoS)","issue":"7","license":[{"start":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T00:00:00Z","timestamp":1783641600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>\n                    Continuous respiratory monitoring enables early detection of physiological deterioration, yet conventional capnography remains impractical for prolonged use. Photoplethysmography (PPG) offers a non-invasive alternative that encodes respiratory information through baseline wander (respiratory-induced intensity variation; RIIV), amplitude modulation (respiratory-induced amplitude variation; RIAV), and frequency modulation (respiratory-induced frequency variation; RIFV) of the pulsatile waveform. Existing PPG-based deep learning approaches, whether operating on the raw signal or on these physiological modulations, are limited to single-task architectures that estimate either respiratory rate or reconstruct the respiratory waveform in isolation, without jointly addressing both outputs. We propose a late fusion multi-task framework in which dedicated encoder branches independently process each modulation before fusion, and dual decoders simultaneously reconstruct the respiratory waveform and estimate the respiratory rate. The framework was evaluated on the CapnoBase (\n                    <jats:italic>n<\/jats:italic>\n                    \u2009=\u200942) and BIDMC (\n                    <jats:italic>n<\/jats:italic>\n                    \u2009=\u200952) benchmarks across multiple training strategies. For respiratory-rate estimation, the best transfer-learning configurations achieved a mean absolute error (MAE) of 2.27 bpm on CapnoBase and 1.33 bpm on BIDMC. For waveform reconstruction, the corresponding MAE values were 19.00% and 20.90%, with moderate correlations (\n                    <jats:italic>r<\/jats:italic>\n                    \u2009=\u20090.662 and\n                    <jats:italic>r<\/jats:italic>\n                    \u2009=\u20090.591, respectively). Sequential transfer learning consistently outperformed all other strategies, whereas pooled training degraded both outputs, demonstrating that capnography-derived and impedance-derived waveforms are not interchangeable training targets. These findings establish that short-window PPG can simultaneously support respiratory-rate estimation and waveform reconstruction, when reference signal compatibility is explicitly addressed in multi-task training.\n                  <\/jats:p>","DOI":"10.1371\/journal.pone.0353203","type":"journal-article","created":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T17:30:56Z","timestamp":1783704656000},"page":"e0353203","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["A late fusion multi-task learning for respiratory waveform and rate estimation from photoplethysmography"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0009-0003-2204-6990","authenticated-orcid":true,"given":"Minh Nhut","family":"Ho","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0705-7002","authenticated-orcid":true,"given":"Kien Trong","family":"Nguyen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,7,10]]},"reference":[{"issue":"21","key":"pone.0353203.ref001","doi-asserted-by":"crossref","first-page":"6396","DOI":"10.3390\/s20216396","article-title":"The importance of respiratory rate monitoring: from healthcare to sport and exercise","volume":"20","author":"A Nicol\u00f2","year":"2020","journal-title":"Sensors (Basel)"},{"issue":"7","key":"pone.0353203.ref002","doi-asserted-by":"crossref","first-page":"354","DOI":"10.1007\/BF02600071","article-title":"Respiratory rate predicts cardiopulmonary arrest for internal medicine inpatients","volume":"8","author":"JF Fieselmann","year":"1993","journal-title":"J Gen Intern Med"},{"key":"pone.0353203.ref003","doi-asserted-by":"crossref","first-page":"80","DOI":"10.1186\/s13049-016-0273-9","article-title":"National Early Warning Score (NEWS) as an emergency department predictor of disease severity and 90-day survival in the acutely dyspneic patient - a prospective observational study","volume":"24","author":"B Bilben","year":"2016","journal-title":"Scand J Trauma Resusc Emerg Med"},{"issue":"1","key":"pone.0353203.ref004","doi-asserted-by":"crossref","first-page":"11","DOI":"10.1186\/s13613-019-0485-z","article-title":"Photoplethysmographic determination of the respiratory rate in acutely ill patients: validation of a new algorithm and implementation into a biomedical device","volume":"9","author":"E L\u2019Her","year":"2019","journal-title":"Ann Intensive Care"},{"issue":"4","key":"pone.0353203.ref005","doi-asserted-by":"crossref","first-page":"610","DOI":"10.1088\/0967-3334\/37\/4\/610","article-title":"An assessment of algorithms to estimate respiratory rate from the electrocardiogram and photoplethysmogram","volume":"37","author":"PH Charlton","year":"2016","journal-title":"Physiol Meas"},{"key":"pone.0353203.ref006","doi-asserted-by":"crossref","DOI":"10.2196\/66347","article-title":"Impact on patient outcomes of continuous vital sign monitoring on medical wards: propensity-matched analysis","volume":"27","author":"B Rowland","year":"2025","journal-title":"J Med Internet Res"},{"issue":"1","key":"pone.0353203.ref007","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1007\/s10916-023-01934-3","article-title":"The effect of continuous versus periodic vital sign monitoring on disease severity of patients with an unplanned ICU transfer","volume":"47","author":"Y Eddahchouri","year":"2023","journal-title":"J Med Syst"},{"issue":"3","key":"pone.0353203.ref008","doi-asserted-by":"crossref","DOI":"10.1088\/0967-3334\/28\/3\/R01","article-title":"Photoplethysmography and its application in clinical physiological measurement","volume":"28","author":"J Allen","year":"2007","journal-title":"Physiol Meas"},{"key":"pone.0353203.ref009","doi-asserted-by":"crossref","first-page":"808451","DOI":"10.3389\/fphys.2021.808451","article-title":"Photoplethysmogram analysis and applications: an integrative review","volume":"12","author":"J Park","year":"2022","journal-title":"Front Physiol"},{"key":"pone.0353203.ref010","doi-asserted-by":"crossref","first-page":"103746","DOI":"10.1016\/j.bspc.2022.103746","article-title":"Realtime PPG based respiration rate estimation for remote health monitoring applications","volume":"77","author":"K Selvakumar","year":"2022","journal-title":"Biomed Signal Process Control"},{"issue":"2","key":"pone.0353203.ref011","doi-asserted-by":"crossref","first-page":"025008","DOI":"10.1088\/1361-6579\/ada9c1","article-title":"Generative adversarial networks with fully connected layers to denoise PPG signals","volume":"46","author":"IA Avila Castro","year":"2025","journal-title":"Physiol Meas"},{"issue":"4","key":"pone.0353203.ref012","doi-asserted-by":"crossref","first-page":"21","DOI":"10.3390\/bioengineering3040021","article-title":"Optimal signal quality index for photoplethysmogram signals","volume":"3","author":"M Elgendi","year":"2016","journal-title":"Bioengineering (Basel)"},{"key":"pone.0353203.ref013","doi-asserted-by":"crossref","first-page":"50","DOI":"10.1186\/1475-925X-13-50","article-title":"Motion artifact removal from photoplethysmographic signals by combining temporally constrained independent component analysis and adaptive filter","volume":"13","author":"F Peng","year":"2014","journal-title":"Biomed Eng Online"},{"key":"pone.0353203.ref014","doi-asserted-by":"crossref","first-page":"2","DOI":"10.1109\/RBME.2017.2763681","article-title":"Breathing rate estimation from the electrocardiogram and photoplethysmogram: a review","volume":"11","author":"PH Charlton","year":"2018","journal-title":"IEEE Rev Biomed Eng"},{"issue":"4","key":"pone.0353203.ref015","doi-asserted-by":"crossref","first-page":"969","DOI":"10.1109\/JBHI.2020.3013811","article-title":"Modulation model of the photoplethysmography signal for vital sign extraction","volume":"25","author":"M Chen","year":"2021","journal-title":"IEEE J Biomed Health Inform"},{"issue":"7","key":"pone.0353203.ref016","doi-asserted-by":"crossref","first-page":"1946","DOI":"10.1109\/TBME.2013.2246160","article-title":"Multiparameter respiratory rate estimation from the photoplethysmogram","volume":"60","author":"W Karlen","year":"2013","journal-title":"IEEE Trans Biomed Eng"},{"key":"pone.0353203.ref017","doi-asserted-by":"crossref","first-page":"701","DOI":"10.3389\/fphys.2017.00701","article-title":"How nonlinear-type time-frequency analysis can help in sensing instantaneous heart rate and instantaneous respiratory rate from photoplethysmography in a reliable way","volume":"8","author":"A Cicone","year":"2017","journal-title":"Front Physiol"},{"issue":"1","key":"pone.0353203.ref018","doi-asserted-by":"crossref","first-page":"1","DOI":"10.3109\/03091902.2011.638965","article-title":"Photoplethysmographic derivation of respiratory rate: a review of relevant physiology","volume":"36","author":"DJ Meredith","year":"2012","journal-title":"J Med Eng Technol"},{"issue":"2","key":"pone.0353203.ref019","doi-asserted-by":"crossref","first-page":"2236","DOI":"10.3390\/s120202236","article-title":"Respiratory physiology and the impact of different modes of ventilation on the photoplethysmographic waveform","volume":"12","author":"AA Alian","year":"2012","journal-title":"Sensors (Basel)"},{"issue":"1","key":"pone.0353203.ref020","doi-asserted-by":"crossref","first-page":"48","DOI":"10.1007\/BF02513265","article-title":"Estimation of respiratory volumes from the photoplethysmographic signal. Part 2: a model study","volume":"37","author":"A Johansson","year":"1999","journal-title":"Med Biol Eng Comput"},{"key":"pone.0353203.ref021","doi-asserted-by":"crossref","first-page":"460","DOI":"10.3389\/fphys.2016.00460","article-title":"Modulations of heart rate, ECG, and cardio-respiratory coupling observed in polysomnography","volume":"7","author":"T Penzel","year":"2016","journal-title":"Front Physiol"},{"issue":"1","key":"pone.0353203.ref022","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1016\/S0008-6363(02)00851-9","article-title":"Hypothesis: respiratory sinus arrhythmia is an intrinsic resting function of cardiopulmonary system","volume":"58","author":"J Hayano","year":"2003","journal-title":"Cardiovasc Res"},{"key":"pone.0353203.ref023","doi-asserted-by":"crossref","first-page":"948","DOI":"10.3389\/fphys.2018.00948","article-title":"Extracting instantaneous respiratory rate from multiple photoplethysmogram respiratory-induced variations","volume":"9","author":"P Dehkordi","year":"2018","journal-title":"Front Physiol"},{"issue":"6","key":"pone.0353203.ref024","doi-asserted-by":"crossref","first-page":"1693","DOI":"10.3390\/s18061693","article-title":"Acquiring respiration rate from photoplethysmographic signal by recursive Bayesian tracking of intrinsic modes in time-frequency spectra","volume":"18","author":"M Pirhonen","year":"2018","journal-title":"Sensors (Basel)"},{"key":"pone.0353203.ref025","doi-asserted-by":"crossref","first-page":"732","DOI":"10.3389\/fphys.2019.00732","article-title":"Toward accurate extraction of respiratory frequency from the photoplethysmogram: effect of measurement site","volume":"10","author":"V Hartmann","year":"2019","journal-title":"Front Physiol"},{"issue":"11","key":"pone.0353203.ref026","doi-asserted-by":"crossref","first-page":"3705","DOI":"10.3390\/s18113705","article-title":"Validation of instantaneous respiratory rate using reflectance PPG from different body positions","volume":"18","author":"D Jarchi","year":"2018","journal-title":"Sensors (Basel)"},{"issue":"3","key":"pone.0353203.ref027","doi-asserted-by":"crossref","first-page":"766","DOI":"10.1109\/JBHI.2017.2679108","article-title":"Ensemble empirical mode decomposition with principal component analysis: a novel approach for extracting respiratory rate and heart rate from photoplethysmographic signal","volume":"22","author":"MA Motin","year":"2018","journal-title":"IEEE J Biomed Health Inform"},{"issue":"11","key":"pone.0353203.ref028","doi-asserted-by":"crossref","first-page":"3238","DOI":"10.3390\/s20113238","article-title":"Estimation of heart rate and respiratory rate from PPG signal using complementary ensemble empirical mode decomposition with both independent component analysis and non-negative matrix factorization","volume":"20","author":"R Lei","year":"2020","journal-title":"Sensors (Basel)"},{"issue":"13","key":"pone.0353203.ref029","doi-asserted-by":"crossref","first-page":"650","DOI":"10.1049\/el.2020.0566","article-title":"Extraction of respiratory rate from PPG using ensemble empirical mode decomposition with Kalman filter","volume":"56","author":"H Sharma","year":"2020","journal-title":"Electron Lett"},{"key":"pone.0353203.ref030","doi-asserted-by":"crossref","first-page":"138","DOI":"10.1016\/j.bspc.2017.03.009","article-title":"Wavelet-based embedded algorithm for respiratory rate estimation from PPG signal","volume":"36","author":"Y-D Lin","year":"2017","journal-title":"Biomed Signal Process Control"},{"key":"pone.0353203.ref031","doi-asserted-by":"crossref","first-page":"107006","DOI":"10.1016\/j.bspc.2024.107006","article-title":"Respiratory rate estimation from photoplethysmogram baseline wandering by harmonic analysis and sequential fusion","volume":"100","author":"C Zhang","year":"2025","journal-title":"Biomed Signal Process Control"},{"issue":"8","key":"pone.0353203.ref032","doi-asserted-by":"crossref","first-page":"1914","DOI":"10.1109\/TBME.2016.2613124","article-title":"Toward a robust estimation of respiratory rate from pulse oximeters","volume":"64","author":"MAF Pimentel","year":"2017","journal-title":"IEEE Trans Biomed Eng"},{"key":"pone.0353203.ref033","doi-asserted-by":"crossref","first-page":"926305","DOI":"10.1155\/2010\/926305","article-title":"Data fusion for improved respiration rate estimation","volume":"2010","author":"S Nemati","year":"2010","journal-title":"EURASIP J Adv Signal Process"},{"issue":"1","key":"pone.0353203.ref034","doi-asserted-by":"crossref","first-page":"170","DOI":"10.1186\/1475-925X-13-170","article-title":"Real-time estimation of respiratory rate from a photoplethysmogram using an adaptive lattice notch filter","volume":"13","author":"C Park","year":"2014","journal-title":"Biomed Eng Online"},{"key":"pone.0353203.ref035","first-page":"5948","article-title":"Respiratory rate estimation using ppg: a deep learning approach","author":"D Bian","year":"2020","journal-title":"Annu Int Conf IEEE Eng Med Biol Soc"},{"key":"pone.0353203.ref036","doi-asserted-by":"crossref","unstructured":"Lampier LC, Coelho YL, Caldeira EMO, Bastos-Filho T. A deep learning approach to estimate the respiratory rate from photoplethysmogram. Ingenius. 2022;(27):96\u2013104. https:\/\/doi.org\/10.17163\/ings.n27.2022.09","DOI":"10.17163\/ings.n27.2022.09"},{"issue":"10","key":"pone.0353203.ref037","doi-asserted-by":"crossref","first-page":"558","DOI":"10.3390\/bioengineering9100558","article-title":"Lightweight end-to-end deep learning solution for estimating the respiration rate from photoplethysmogram signal","volume":"9","author":"MH Chowdhury","year":"2022","journal-title":"Bioengineering (Basel)"},{"key":"pone.0353203.ref038","doi-asserted-by":"crossref","first-page":"105338","DOI":"10.1016\/j.compbiomed.2022.105338","article-title":"Deep learning for predicting respiratory rate from biosignals","volume":"144","author":"AK Kumar","year":"2022","journal-title":"Comput Biol Med"},{"key":"pone.0353203.ref039","doi-asserted-by":"crossref","first-page":"104992","DOI":"10.1016\/j.bspc.2023.104992","article-title":"Rapid extraction of respiratory waveforms from photoplethysmography: a deep corr-encoder approach","volume":"85","author":"HJ Davies","year":"2023","journal-title":"Biomed Signal Process Control"},{"issue":"4","key":"pone.0353203.ref040","doi-asserted-by":"crossref","first-page":"1705","DOI":"10.1007\/s13246-024-01482-1","article-title":"PPG2RespNet: a deep learning model for respirational signal synthesis and monitoring from photoplethysmography (PPG) signal","volume":"47","author":"MNI Shuzan","year":"2024","journal-title":"Phys Eng Sci Med"},{"issue":"18","key":"pone.0353203.ref041","doi-asserted-by":"crossref","first-page":"15943","DOI":"10.1109\/JIOT.2023.3265980","article-title":"RRWaveNet: a compact end-to-end multiscale residual CNN for robust PPG respiratory rate estimation","volume":"10","author":"P Osathitporn","year":"2023","journal-title":"IEEE Internet Things J"},{"key":"pone.0353203.ref042","doi-asserted-by":"crossref","first-page":"637","DOI":"10.1109\/OJEMB.2023.3329728","article-title":"Continuous patient-independent estimation of respiratory rate and blood pressure using robust spectro-temporal features derived from photoplethysmogram only","volume":"5","author":"MA Sultan","year":"2023","journal-title":"IEEE Open J Eng Med Biol"},{"key":"pone.0353203.ref043","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/TIM.2020.3043506","article-title":"i-PRExT: photoplethysmography derived respiration signal extraction and respiratory rate tracking using neural networks","volume":"70","author":"B Roy","year":"2021","journal-title":"IEEE Trans Instrum Meas"},{"issue":"3","key":"pone.0353203.ref044","doi-asserted-by":"crossref","first-page":"284","DOI":"10.3390\/diagnostics14030284","article-title":"A novel respiratory rate estimation algorithm from photoplethysmogram using deep learning model","volume":"14","author":"WJ Chin","year":"2024","journal-title":"Diagnostics (Basel)"},{"issue":"12","key":"pone.0353203.ref045","doi-asserted-by":"crossref","first-page":"3980","DOI":"10.3390\/s24123980","article-title":"Energy-efficient PPG-based respiratory rate estimation using spiking neural networks","volume":"24","author":"G Yang","year":"2024","journal-title":"Sensors (Basel)"},{"key":"pone.0353203.ref046","doi-asserted-by":"crossref","first-page":"1046","DOI":"10.1007\/s11227-025-07543-w","article-title":"FPGA-based SqueezeNet transformer with multi-scale attention fusion for real-time respiration rate estimation using PPG signals","volume":"81","author":"IJ Selvam","year":"2025","journal-title":"J Supercomput"},{"issue":"9","key":"pone.0353203.ref047","doi-asserted-by":"crossref","first-page":"1024","DOI":"10.3390\/bioengineering10091024","article-title":"Predicting respiratory rate from electrocardiogram and photoplethysmogram using a transformer-based model","volume":"10","author":"Q Zhao","year":"2023","journal-title":"Bioengineering (Basel)"},{"issue":"9","key":"pone.0353203.ref048","doi-asserted-by":"crossref","first-page":"2033","DOI":"10.1109\/TBME.2017.2778265","article-title":"A robust fusion model for estimating respiratory rate from photoplethysmography and electrocardiography","volume":"65","author":"DA Birrenkott","year":"2018","journal-title":"IEEE Trans Biomed Eng"},{"issue":"10","key":"pone.0353203.ref049","doi-asserted-by":"crossref","first-page":"1222","DOI":"10.3390\/bioengineering10101222","article-title":"Evaluation of the photoplethysmogram-based deep learning model for continuous respiratory rate estimation in surgical intensive care unit","volume":"10","author":"CS Hwang","year":"2023","journal-title":"Bioengineering (Basel)"},{"key":"pone.0353203.ref050","doi-asserted-by":"crossref","first-page":"109798","DOI":"10.1016\/j.compbiomed.2025.109798","article-title":"Multitask learning approach for PPG applications: case studies on signal quality assessment and physiological parameters estimation","volume":"188","author":"M Feli","year":"2025","journal-title":"Comput Biol Med"},{"key":"pone.0353203.ref051","first-page":"25","article-title":"CapnoBase: Signal database and tools to collect, share and annotate respiratory signals.","volume-title":"2010 Annual Meeting of the Society for Technology in Anesthesia","author":"W Karlen","year":"2010"},{"issue":"23","key":"pone.0353203.ref052","doi-asserted-by":"crossref","DOI":"10.1161\/01.CIR.101.23.e215","article-title":"PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals","volume":"101","author":"AL Goldberger","year":"2000","journal-title":"Circulation"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353203","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T17:31:04Z","timestamp":1783704664000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353203"}},"subtitle":[],"editor":[{"given":"Haipeng","family":"Liu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,7,10]]},"references-count":52,"journal-issue":{"issue":"7","published-online":{"date-parts":[[2026,7,10]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0353203","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,10]]}}}