{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,18]],"date-time":"2026-05-18T06:07:49Z","timestamp":1779084469155,"version":"3.51.4"},"reference-count":41,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,5,18]],"date-time":"2026-05-18T00:00:00Z","timestamp":1779062400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Neurosci."],"abstract":"\n Background and objectives<\/jats:title>\n Elderly patients (\u226565 years) who sustain burn injuries encounter a clinically significant perioperative challenge: a dysregulated hyperinflammatory response, characterized by elevated levels of interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-\u03b1), and C-reactive protein (CRP), compounded by a markedly reduced hemodynamic reserve. Both propofol and low-dose ketamine exhibit distinct anti-inflammatory mechanisms; however, the optimization of their combined dosing within explicit safety parameters remains unestablished. Our objectives were to: (1) develop and externally validate a probabilistic machine learning (ML) model to predict dynamic 24-h trajectories of inflammatory markers; and (2) integrate these predictions with a safety-constrained offline reinforcement learning (RL) agent to formulate individualized propofol-ketamine dosing recommendations.<\/jats:p>\n <\/jats:sec>\n \n Study design<\/jats:title>\n This study employed a retrospective multi-cohort analysis utilizing two publicly accessible intensive care databases.<\/jats:p>\n <\/jats:sec>\n \n Setting<\/jats:title>\n The research was conducted in an academic medical center ICU (MIMIC-IV) and across 208 community and academic hospitals (eICU Collaborative Research Database).<\/jats:p>\n <\/jats:sec>\n \n Measurements<\/jats:title>\n The study analyzed 614 perioperative episodes in patients aged \u226565 years with confirmed burn injuries who received propofol-based anesthesia for \u226530 min and had \u22652 inflammatory laboratory measurements within 6\u201324 h post-induction. External validation was performed on 206 independent episodes.<\/jats:p>\n <\/jats:sec>\n \n Main results<\/jats:title>\n \n The proposed Event-Transformer with continuous-time Neural ODE dynamics demonstrated a 12-h IL-6 mean absolute error (MAE) of 6.82 pg\/mL, representing a 70.1% improvement over linear mixed models (22.8 pg\/mL). It achieved an inflammatory spike detection area under the receiver operating characteristic curve (AUROC) of 0.814 and empirical 90% prediction interval (PI) coverage of 87.2%. The Conservative Policy with Q-Learning (CPQL) dosing agent enhanced the time within the MAP target range (65\u201390 mmHg) from 62.3% to 71.8% (\n p<\/jats:italic>\n &lt; 0.001), decreased vasopressor initiation from 27.0% to 18.4% (\n p<\/jats:italic>\n = 0.003), reduced peak predicted CRP by 21.3%, and decreased total propofol exposure by 12.1% through the introduction of adjunct ketamine (\u22487.2 mcg\/kg\/min). The safety constraint violation rate was 0.0% under CPQL compared to 4.2% for unconstrained offline RL.\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n An integrated inflammatory forecasting and dosing optimization pipeline can facilitate individualized propofol-ketamine titration in elderly burn patients, yielding predicted clinically significant improvements in hemodynamic stability and inflammatory burden, without safety violations. Clinically, the 70.1% reduction in IL-6 forecasting error translates to a meaningful difference between correct and incorrect inflammatory spike classification in a substantial fraction of patients, supporting the potential real-world utility of this framework as a decision-support tool to inform and guide future prospective trials.<\/jats:p>\n <\/jats:sec>","DOI":"10.3389\/fncom.2026.1824898","type":"journal-article","created":{"date-parts":[[2026,5,18]],"date-time":"2026-05-18T05:42:57Z","timestamp":1779082977000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Deep learning guided propofol ketamine dosing and inflammation trajectories in elderly burns"],"prefix":"10.3389","volume":"20","author":[{"given":"Xiaohui","family":"Yuan","sequence":"first","affiliation":[{"name":"Department of Anesthesiology, Wuhan Third Hospital","place":["Wuhan, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Gang","family":"Wang","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology, Wuhan Third Hospital","place":["Wuhan, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xiaoyang","family":"Jiang","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology, Wuhan Third Hospital","place":["Wuhan, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenjing","family":"Miao","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology, Wuhan Third Hospital","place":["Wuhan, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2026,5,18]]},"reference":[{"key":"B1","unstructured":"Bennett\n N.\n \n \n Ple\u010dko\n D.\n \n \n Ukor\n I.-F.\n \n \n eth-mds\/ricu: Icu data with R"},{"key":"B2","doi-asserted-by":"publisher","first-page":"giad041","DOI":"10.1093\/gigascience\/giad041","article-title":"ricu: R's interface to intensive care data","volume":"12","author":"Bennett","year":"","journal-title":"GigaScience"},{"key":"B3","article-title":"\u201cNeural ordinary differential equations,\u201d","author":"Chen","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B4","doi-asserted-by":"publisher","first-page":"107468","DOI":"10.1016\/j.burns.2025.107468","article-title":"Redefining the concept of the elderly burn patient: analysis of a multicentre international dataset","volume":"51","author":"Dempsey","year":"2025","journal-title":"Burns"},{"key":"B5","doi-asserted-by":"publisher","first-page":"840","DOI":"10.1097\/TA.0000000000000124","article-title":"External validation of the revised baux score for the prediction of mortality in patients with acute burn injury","volume":"76","author":"Dokter","year":"2014","journal-title":"J. Trauma Acute Care Surg"},{"key":"B6","doi-asserted-by":"publisher","first-page":"13","DOI":"10.1097\/01.shk.0000223120.26394.7d","article-title":"Cytokine expression profile over time in severely burned pediatric patients","volume":"26","author":"Finnerty","year":"2006","journal-title":"Shock"},{"key":"B7","doi-asserted-by":"publisher","first-page":"78","DOI":"10.1016\/j.exger.2017.10.022","article-title":"Age-related immune responses after burn and inhalation injury are associated with altered clinical outcomes","volume":"105","author":"Frankel","year":"2018","journal-title":"Exp. Gerontol"},{"key":"B8","doi-asserted-by":"publisher","first-page":"R81","DOI":"10.1186\/cc6932","article-title":"Are serum cytokines early predictors for the outcome of burn patients with inhalation injuries who do not survive?","volume":"12","author":"Gauglitz","year":"2008","journal-title":"Critical Care"},{"key":"B9","doi-asserted-by":"publisher","first-page":"tkaa027","DOI":"10.1093\/burnst\/tkaa027","article-title":"Burns in the elderly: a nationwide study on management and clinical outcomes","volume":"8","author":"Goei","year":"2020","journal-title":"Burns Trauma"},{"key":"B10","doi-asserted-by":"publisher","first-page":"16","DOI":"10.1038\/s41591-018-0310-5","article-title":"Guidelines for reinforcement learning in healthcare","volume":"25","author":"Gottesman","year":"2019","journal-title":"Nat. Med"},{"key":"B11","article-title":"Gaussian error linear units (gelus)","author":"Hendrycks","year":"2016","journal-title":"arXiv preprint arXiv:1606.08415"},{"key":"B12","doi-asserted-by":"publisher","first-page":"e17598","DOI":"10.1371\/journal.pone.0017598","article-title":"Anesthetic propofol reduces endotoxic inflammation by inhibiting reactive oxygen species-regulated akt\/ikk\u03b2\/nf-\u03bab signaling","volume":"6","author":"Hsing","year":"2011","journal-title":"PLoS ONE"},{"key":"B13","doi-asserted-by":"publisher","first-page":"S239","DOI":"10.1097\/01.CCM.0000232625.63460.D4","article-title":"Advances in burn critical care","volume":"34","author":"Ipaktchi","year":"2006","journal-title":"Crit. Care Med"},{"key":"B14","doi-asserted-by":"publisher","first-page":"638","DOI":"10.1111\/j.1399-6576.2011.02446.x","article-title":"Ketamine as an analgesic in the pre-hospital setting: a systematic review","volume":"55","author":"Jennings","year":"2011","journal-title":"Acta Anaesthesiol. Scand"},{"key":"B15","doi-asserted-by":"publisher","first-page":"11","DOI":"10.1038\/s41572-020-0145-5","article-title":"Burn injury","volume":"6","author":"Jeschke","year":"2020","journal-title":"Nat. Rev. Dis. Prim"},{"key":"B16","unstructured":"Johnson\n A. E. W.\n \n \n MIMIC-IV Clinical Database (version 2.2)\n \n 2022"},{"key":"B17","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/s41597-023-01945-2","article-title":"MIMIC-IV, a freely accessible electronic health record dataset","volume":"10","author":"Johnson","year":"2023","journal-title":"Sci. Data"},{"key":"B18","unstructured":"Kidger\n P.\n \n \n On neural differential equations\n \n 2022"},{"key":"B19","doi-asserted-by":"publisher","first-page":"66","DOI":"10.1097\/BCR.0b013e318203336a","article-title":"Functional and psychosocial outcomes of older adults after burn injury: results from a multicenter database of severe burn injury","volume":"32","author":"Klein","year":"2011","journal-title":"J. Burn Care Res"},{"key":"B20","first-page":"1179","article-title":"\u201cConservative q-learning for offline reinforcement learning,\u201d","author":"Kumar","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B21","first-page":"403","article-title":"\u201cDuett: dual event time transformer for electronic health records,\u201d","volume-title":"Machine Learning for Healthcare Conference","author":"Labach","year":"2023"},{"key":"B22","article-title":"Offline reinforcement learning: tutorial, review, and perspectives on open problems","author":"Levine","year":"2020","journal-title":"arXiv preprint arXiv:2005.01643"},{"key":"B23","doi-asserted-by":"publisher","first-page":"S715","DOI":"10.1016\/S0003-4975(02)04701-X","article-title":"Inflammatory response to cardiopulmonary bypass","volume":"75","author":"Levy","year":"2003","journal-title":"Ann. Thorac. Surg"},{"key":"B24","doi-asserted-by":"publisher","first-page":"776","DOI":"10.1097\/BCR.0b013e3181b47ed2","article-title":"The effect of burn center and burn center volume on the mortality of burned adults\u2014an analysis of the data in the national burn repository","volume":"30","author":"Light","year":"2009","journal-title":"J. Burn Care Res"},{"key":"B25","doi-asserted-by":"publisher","first-page":"93","DOI":"10.1016\/j.burns.2023.06.010","article-title":"Development and validation of a nomogram to predict hypothermia in adult burn patients during escharectomy under general anesthesia","volume":"50","author":"Lou","year":"2024","journal-title":"Burns"},{"key":"B26","doi-asserted-by":"publisher","first-page":"1603104","DOI":"10.3389\/fcimb.2025.1603104","article-title":"The j-shaped association between the ratio of neutrophil counts to prognostic nutritional index and mortality in icu patients with sepsis: a retrospective study based on the MIMIC database","volume":"15","author":"Lou","year":"","journal-title":"Front. Cell. Infect. Microbiol"},{"key":"B27","doi-asserted-by":"publisher","first-page":"2537917","DOI":"10.1080\/07853890.2025.2537917","article-title":"Trends and levels of the global, regional, and national burden of injuries from 1990 to 2021: findings from the global burden of disease study 2021","volume":"57","author":"Lou","year":"","journal-title":"Ann. Med"},{"key":"B28","doi-asserted-by":"publisher","first-page":"1555082","DOI":"10.3389\/fendo.2025.1555082","article-title":"Association between serum glucose potassium ratio and short- and long-term all-cause mortality in patients with sepsis admitted to the intensive care unit: a retrospective analysis based on the MIMIC-IV database","volume":"16","author":"Lou","year":"","journal-title":"Front. Endocrinol"},{"key":"B29","doi-asserted-by":"publisher","first-page":"1551","DOI":"10.1111\/j.1526-4637.2011.01220.x","article-title":"A systematic review of ketamine as an analgesic agent in adult burn injuries","volume":"12","author":"McGuinness","year":"2011","journal-title":"Pain Med"},{"key":"B30","doi-asserted-by":"publisher","first-page":"e220004","DOI":"10.20900\/agmr20220004","article-title":"Aging, cutaneous burn injury and multi-organ complications: the role of the gut","volume":"4","author":"McMahan","year":"2022","journal-title":"Adv. Geriatr. Med. Res"},{"key":"B31","doi-asserted-by":"publisher","first-page":"690","DOI":"10.1097\/TA.0b013e3181c453b3","article-title":"Simplified estimates of the probability of death after burn injuries: extending and updating the baux score","volume":"68","author":"Osler","year":"2010","journal-title":"J. Trauma Acute Care Surg"},{"key":"B32","doi-asserted-by":"publisher","first-page":"180178","DOI":"10.1038\/sdata.2018.178","article-title":"The eicu collaborative research database, a freely available multi-center database for critical care research","volume":"5","author":"Pollard","year":"2018","journal-title":"Sci. Data"},{"key":"B33","unstructured":"Pollard\n T. J.\n \n \n Johnson\n A. E. W.\n \n \n eICU Collaborative Research Database (version 2.0)\n \n 2019"},{"key":"B34","doi-asserted-by":"publisher","first-page":"31","DOI":"10.1038\/s41591-021-01614-0","article-title":"AI in health and medicine","volume":"28","author":"Rajpurkar","year":"2022","journal-title":"Nat. Med"},{"key":"B35","article-title":"\u201cLatent ordinary differential equations for irregularly-sampled time series,\u201d","author":"Rubanova","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B36","unstructured":"Dropout: a simple way to prevent neural networks from overfitting\n \n 1929\n 1958\n \n \n Srivastava\n N.\n \n \n Hinton\n G.\n \n \n Krizhevsky\n A.\n \n \n Sutskever\n I.\n \n \n Salakhutdinov\n R.\n \n \n J. Mach. Learn. Res\n 15\n 2014"},{"key":"B37","doi-asserted-by":"publisher","first-page":"195","DOI":"10.1097\/SLA.0000000000001408","article-title":"Impaired immune response in elderly burn patients: new insights into the immune-senescence phenotype","volume":"264","author":"Stanojcic","year":"2016","journal-title":"Ann. Surg"},{"key":"B38","doi-asserted-by":"publisher","first-page":"a016295","DOI":"10.1101\/cshperspect.a016295","article-title":"Il-6 in inflammation, immunity, and disease","volume":"6","author":"Tanaka","year":"2014","journal-title":"Cold Spring Harb. Perspect. Biol"},{"key":"B39","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3516367","article-title":"Self-supervised transformer for sparse and irregularly sampled multivariate clinical time-series","volume":"16","author":"Tipirneni","year":"2022","journal-title":"ACM Trans. Knowl. Disc. Data"},{"key":"B40","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.ejphar.2009.01.007","article-title":"Propofol: a review of its non-anaesthetic effects","volume":"605","author":"Vasileiou","year":"2009","journal-title":"Eur. J. Pharmacol"},{"key":"B41","article-title":"\u201cAttention is all you need,\u201d","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"}],"container-title":["Frontiers in Computational Neuroscience"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fncom.2026.1824898\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,18]],"date-time":"2026-05-18T05:43:00Z","timestamp":1779082980000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fncom.2026.1824898\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,18]]},"references-count":41,"alternative-id":["10.3389\/fncom.2026.1824898"],"URL":"https:\/\/doi.org\/10.3389\/fncom.2026.1824898","relation":{},"ISSN":["1662-5188"],"issn-type":[{"value":"1662-5188","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,18]]},"article-number":"1824898"}}