{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,19]],"date-time":"2026-11-19T21:44:33Z","timestamp":1795124673365,"version":"4.10.3"},"reference-count":31,"publisher":"BMJ","issue":"1","license":[{"start":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T00:00:00Z","timestamp":1772064000000},"content-version":"unspecified","delay-in-days":25,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":["bmj.com"],"crossmark-restriction":true},"short-container-title":["BMJ Health Care Inform"],"accepted":{"date-parts":[[2025,12,27]]},"published-print":{"date-parts":[[2026,2]]},"abstract":"<jats:sec>\n                    <jats:title>Objectives<\/jats:title>\n                    <jats:p>Predicting mortality is vital for tailoring treatments, improving care and reducing costs. Machine learning (ML) has shown strong potential, often outperforming traditional severity-of-illness scoring systems in intensive care units (ICUs). However, the black-box nature of ML limits adoption. This study evaluates the accuracy of several ML algorithms on the Medical Information Mart for Intensive Care III (MIMIC-III) dataset and applies explainable artificial intelligence to identify variables influencing ICU mortality.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>A retrospective cohort of 600 MIMIC-III patient records was analysed. ML algorithms tested included support vector machine (SVM), K-nearest neighbour (KNN), decision trees (DT), gradient boosting (GB), random forests (RF), Naive Bayes (NB), logistic regression (LR) and extra trees (ET). Models were assessed with threefold cross-validation using F1 Score, sensitivity, specificity, confusion matrix and accuracy. SHapley Additive exPlanations (SHAP) was applied to explain key factors influencing predictions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Among the eight algorithms evaluated, ET and GB achieved the highest accuracy (98.33% and 98.23%, respectively) with F1-scores above 96%. SVM also performed strongly (97.50% accuracy, F1=94.34%). RF and DT yielded accuracies of 95.00% and 94.17%, respectively. NB reached 96.67% accuracy with 100% recall, while KNN and LR showed lower discriminative performance (76.67% and 75.83%\u2009accuracy, respectively).<\/jats:p>\n                    <jats:p>SHAP, LIME and Shapley values consistently identified hypertension, tumours and endocrine and digestive disease as the leading predictors of mortality.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>Findings highlight ML\u2019s potential to optimise ICU decision-making and support clinicians. However, reliance on retrospective data remains a limitation, and clinical validation is required.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>ML algorithms are highly effective for mortality prediction, and explainability is key for trust and adoption. When combined, accuracy and interpretability enable ML to safely support informed ICU decisions and improve patient outcomes.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1136\/bmjhci-2024-101406","type":"journal-article","created":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T14:00:10Z","timestamp":1772114410000},"page":"e101406","update-policy":"https:\/\/doi.org\/10.1136\/crossmarkpolicy","source":"Crossref","is-referenced-by-count":2,"title":["Explainable AI for mortality prediction: a comparative study using the MIMIC-III dataset"],"prefix":"10.1136","volume":"33","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7668-8524","authenticated-orcid":false,"given":"Niusha","family":"Shafiabady","sequence":"first","affiliation":[{"name":"Women in AI for Social Good Lab & Discipline of IT, Australian Catholic University, North Sydney, New South Wales, Australia"},{"name":"Charles Darwin University, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dave","family":"Akume","sequence":"additional","affiliation":[{"name":"Charles Darwin University, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-7771-1250","authenticated-orcid":false,"given":"Mohammadreza","family":"Haghighat","sequence":"additional","affiliation":[{"name":"Department of Biomedical Engineering, Amirkabir University of Technology, Tehran, Iran (the Islamic Republic of)"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fareed","family":"Ud Din","sequence":"additional","affiliation":[{"name":"University of New England School of Science and Technology, Armidale, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kabir","family":"Sattarshetty","sequence":"additional","affiliation":[{"name":"Charles Darwin University, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Asif","family":"Karim","sequence":"additional","affiliation":[{"name":"Charles Darwin University, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jianlong","family":"Zhou","sequence":"additional","affiliation":[{"name":"University of Technology Sydney, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ethar","family":"Alsharaydeh","sequence":"additional","affiliation":[{"name":"Western Sydney University, Sydney, New South Wales, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"239","published-online":{"date-parts":[[2026,2,26]]},"reference":[{"key":"2026022606000605000_33.1.e101406.1","doi-asserted-by":"crossref","first-page":"13","DOI":"10.5005\/jp-journals-10071-24063","article-title":"Mortality Prediction in the ICU: The Daunting Task of Predicting the Unpredictable","volume":"26","author":"Kumar","year":"2022","journal-title":"Indian J Crit Care Med"},{"key":"2026022606000605000_33.1.e101406.2","doi-asserted-by":"crossref","DOI":"10.1186\/s13054-019-2564-9","article-title":"Use of machine learning to analyse routinely collected intensive care unit data: a systematic review","volume":"23","author":"Shillan","year":"2019","journal-title":"Crit Care"},{"key":"2026022606000605000_33.1.e101406.3","doi-asserted-by":"crossref","DOI":"10.1016\/j.heliyon.2023.e13200","article-title":"Mortality prediction among ICU inpatients based on MIMIC-III database results from the conditional medical generative adversarial network","volume":"9","author":"Yang","year":"2023","journal-title":"Heliyon"},{"key":"2026022606000605000_33.1.e101406.4","doi-asserted-by":"publisher","DOI":"10.1097\/00003246-198608000-00028"},{"key":"2026022606000605000_33.1.e101406.5","doi-asserted-by":"crossref","first-page":"1280","DOI":"10.1007\/s00134-012-2578-0","article-title":"Comparison between SAPS II and SAPS 3 in predicting hospital mortality in a cohort of 103 Italian ICUs. Is new always better?","volume":"38","author":"Poole","year":"2012","journal-title":"Intensive Care Med"},{"key":"2026022606000605000_33.1.e101406.6","doi-asserted-by":"crossref","DOI":"10.1038\/s41467-021-20910-4","article-title":"Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare","volume":"12","author":"Goh","year":"2021","journal-title":"Nat Commun"},{"key":"2026022606000605000_33.1.e101406.7","doi-asserted-by":"publisher","DOI":"10.1097\/00003246-198411000-00012"},{"key":"2026022606000605000_33.1.e101406.8","doi-asserted-by":"crossref","unstructured":"Sarkar R , Martin C , Mattie H , et al . Performance of intensive care unit severity scoring systems across different ethnicities [Update in: Lancet Digit Health. 2021;3:e241-e249. doi:10.1016\/S2589-7500(21)00022-4]. medRxiv 2021;3. doi:10.1101\/2021.01.19.21249222","DOI":"10.1016\/S2589-7500(21)00022-4"},{"key":"2026022606000605000_33.1.e101406.9","doi-asserted-by":"publisher","DOI":"10.1097\/MCC.0000000000000135"},{"key":"2026022606000605000_33.1.e101406.10","doi-asserted-by":"publisher","DOI":"10.1001\/jama.1993.03510240069035"},{"key":"2026022606000605000_33.1.e101406.11","doi-asserted-by":"publisher","DOI":"10.1097\/00003246-198108000-00008"},{"key":"2026022606000605000_33.1.e101406.12","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0206862","article-title":"Optimal intensive care outcome prediction over time using machine learning","volume":"13","author":"Meiring","year":"2018","journal-title":"PLoS One"},{"key":"2026022606000605000_33.1.e101406.13","doi-asserted-by":"publisher","DOI":"10.1016\/S2589-7500(20)30018-2"},{"key":"2026022606000605000_33.1.e101406.14","doi-asserted-by":"crossref","DOI":"10.21037\/atm-20-6624","article-title":"Mortality prediction for patients with acute respiratory distress syndrome based on machine learning: a population-based study","volume":"9","author":"Huang","year":"2021","journal-title":"Ann Transl Med"},{"key":"2026022606000605000_33.1.e101406.15","doi-asserted-by":"crossref","first-page":"2861","DOI":"10.1016\/j.csbj.2022.06.003","article-title":"Application of interpretable machine learning for early prediction of prognosis in acute kidney injury","volume":"20","author":"Hu","year":"2022","journal-title":"Comput Struct Biotechnol J"},{"key":"2026022606000605000_33.1.e101406.16","doi-asserted-by":"crossref","first-page":"87","DOI":"10.1093\/jamiaopen\/ooy011","article-title":"Predictive modeling in urgent care: a comparative study of machine learning approaches","volume":"1","author":"Tang","year":"2018","journal-title":"JAMIA Open"},{"key":"2026022606000605000_33.1.e101406.17","doi-asserted-by":"crossref","DOI":"10.1038\/sdata.2016.35","article-title":"MIMIC-III, a freely accessible critical care database","volume":"3","author":"Johnson","year":"2016","journal-title":"Sci Data"},{"key":"2026022606000605000_33.1.e101406.18","doi-asserted-by":"crossref","DOI":"10.1038\/s41597-022-01899-x","article-title":"MIMIC-IV, a freely accessible electronic health record dataset","volume":"10","author":"Johnson","year":"2023","journal-title":"Sci Data"},{"key":"2026022606000605000_33.1.e101406.19","doi-asserted-by":"crossref","unstructured":"Tsoni R , Kaldis V , Kapogianni I , et al . A machine learning pipeline using knime to predict hospital admission in the MIMIC-IV database. 2023 14th International Conference on Information, Intelligence, Systems & Applications (IISA); Volos, Greece, 2023:1\u20136. doi:10.1109\/IISA59645.2023.10345903","DOI":"10.1109\/IISA59645.2023.10345903"},{"key":"2026022606000605000_33.1.e101406.20","first-page":"517","article-title":"Automated ML Techniques for Predicting COVID-19 Mortality in the ICU","volume":"305","author":"Sakagianni","year":"2023","journal-title":"Stud Health Technol Inform"},{"key":"2026022606000605000_33.1.e101406.21","doi-asserted-by":"crossref","first-page":"850","DOI":"10.1007\/s00134-010-1772-1","article-title":"The interrater reliability of SAPS II and SAPS 3","volume":"36","author":"Strand","year":"2010","journal-title":"Intensive Care Med"},{"key":"2026022606000605000_33.1.e101406.22","doi-asserted-by":"crossref","DOI":"10.1136\/bmjopen-2018-025925","article-title":"Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK","volume":"9","author":"McWilliams","year":"2019","journal-title":"BMJ Open"},{"key":"2026022606000605000_33.1.e101406.23","doi-asserted-by":"crossref","DOI":"10.1186\/s12911-020-01332-6","article-title":"Explainability for artificial intelligence in healthcare: a multidisciplinary perspective","volume":"20","author":"Amann","year":"2020","journal-title":"BMC Med Inform Decis Mak"},{"key":"2026022606000605000_33.1.e101406.24","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-023-35900-3","article-title":"An FSV analysis approach to verify the robustness of the triple-correlation analysis theoretical framework","volume":"13","author":"Wu","year":"2023","journal-title":"Sci Rep"},{"key":"2026022606000605000_33.1.e101406.25","doi-asserted-by":"crossref","unstructured":"Li Y , Zhou J , Dong Y , et al . ACGAN-gnnexplainer: auxiliary conditional generative explainer for graph neural networks. CIKM \u201923: Proceedings of the 32nd ACM International Conference on Information and Knowledge Management; Birmingham United Kingdom, 2023 doi:10.1145\/3583780.3614772 Available: https:\/\/dl.acm.org\/doi\/proceedings\/10.1145\/3583780","DOI":"10.1145\/3583780.3614772"},{"key":"2026022606000605000_33.1.e101406.26","doi-asserted-by":"publisher","DOI":"10.4103\/sja.SJA_543_18"},{"key":"2026022606000605000_33.1.e101406.27","doi-asserted-by":"crossref","first-page":"2913","DOI":"10.3390\/make6040139","article-title":"Reliable and Faithful Generative Explainers for Graph Neural Networks","volume":"6","author":"Li","year":"2024","journal-title":"MAKE"},{"key":"2026022606000605000_33.1.e101406.28","doi-asserted-by":"crossref","first-page":"1968","DOI":"10.1016\/j.procs.2023.01.173","article-title":"Simplified & Novel Predictive Model using Feature Engineering over MIMIC-III Dataset","volume":"218","author":"Khope","year":"2023","journal-title":"Procedia Comput Sci"},{"key":"2026022606000605000_33.1.e101406.29","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-018-0213-5"},{"key":"2026022606000605000_33.1.e101406.30","doi-asserted-by":"crossref","unstructured":"Rajkomar A , Oren E , Chen K , et al . Scalable and accurate deep learning with electronic health records. NPJ Digit Med 2018;1. doi:10.1038\/s41746-018-0029-1","DOI":"10.1038\/s41746-018-0029-1"},{"key":"2026022606000605000_33.1.e101406.31","doi-asserted-by":"crossref","unstructured":"Raza S , Dolatabadi E , Ondrusek N , et al . Discovering social determinants of health from case reports using natural language processing: algorithmic development and validation. BMC Digit Health 2023;1. doi:10.1186\/s44247-023-00035-y","DOI":"10.1186\/s44247-023-00035-y"}],"container-title":["BMJ Health &amp; Care Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1136\/bmjhci-2024-101406","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T14:00:14Z","timestamp":1772114414000},"score":1,"resource":{"primary":{"URL":"https:\/\/informatics.bmj.com\/lookup\/doi\/10.1136\/bmjhci-2024-101406"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2]]},"references-count":31,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,2,26]]},"published-print":{"date-parts":[[2026,2]]}},"alternative-id":["10.1136\/bmjhci-2024-101406"],"URL":"https:\/\/doi.org\/10.1136\/bmjhci-2024-101406","relation":{},"ISSN":["2632-1009"],"issn-type":[{"value":"2632-1009","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2]]}}}