{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,6]],"date-time":"2026-10-06T11:40:48Z","timestamp":1791286848104,"version":"4.1.0"},"reference-count":41,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T00:00:00Z","timestamp":1766102400000},"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. Public Health"],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Unscheduled return visits (URVs) to emergency departments (EDs) contribute significantly to healthcare burden through resource utilization and ED overcrowding. While artificial intelligence (AI) methodologies show potential in URV prediction, existing studies have employed limited algorithms with moderate performance, highlighting the need for comprehensive AI architecture comparison within unified cohorts.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>This study evaluated the predictive performance of multiple AI models for 72-h ED URVs, aiming to identify optimal risk stratification strategies for improved discharge planning and targeted interventions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This retrospective study analyzed adult internal medicine visits to the ED at a tertiary hospital. URVs were defined as ED revisits occurring within 72\u202fh after initial ED discharge time. The dataset was partitioned into training (70%) and testing (30%) sets. Four traditional machine learning algorithms (logistic regression, support vector machine, random forest, and extreme gradient boosting) and one deep learning architecture (TabNet) were developed with Bayesian optimization for hyperparameter tuning. Model performance was assessed through comprehensive metrics including discrimination, calibration, clinical utility, and confusion matrices. The optimal model underwent feature importance analysis, systematic ablation studies, sensitivity analyses, and subgroup fairness evaluation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Of 143,192 analyzed visits, 24,117 (16.8%) were classified as URVs. Data were allocated into training (\n                      <jats:italic>n<\/jats:italic>\n                      =\u202f100,235) and testing (\n                      <jats:italic>n<\/jats:italic>\n                      =\u202f42,957) sets with consistent URV proportions. TabNet demonstrated optimal discriminative performance with AUROC 0.867 (95% CI: 0.854\u20130.880) and sensitivity of 0.809 (95% CI: 0.801\u20130.816). Decision curve analysis demonstrated sustained clinical utility across threshold probabilities of 10\u201330%. Feature importance analysis identified initial diagnoses of digestive and respiratory system diseases, patient age, P3 triage classification, and ED visit frequency as key predictive variables. Subgroup analysis confirmed consistent performance across patient demographics and clinical characteristics.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>TabNet outperformed traditional machine learning approaches in predicting 72-h ED URVs, offering potential for improved risk stratification in emergency care settings.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fpubh.2025.1609206","type":"journal-article","created":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T06:25:49Z","timestamp":1766125549000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Performance comparison of artificial intelligence models in predicting 72-h emergency department unscheduled return visits"],"prefix":"10.3389","volume":"13","author":[{"given":"Lumin","family":"Fan","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xinghua","family":"Zuo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lunxian","family":"Tang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Honglin","family":"Xiong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yanli","family":"You","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chongjun","family":"Fan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,12,19]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"324","DOI":"10.1136\/emermed-2013-202719","article-title":"Unscheduled return visits (URV) in adults to the emergency department (ED): a rapid evidence assessment policy review","volume":"32","author":"Trivedy","year":"2015","journal-title":"Emerg Med J"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"1074","DOI":"10.1001\/archinte.160.8.1074","article-title":"Hospital readmissions as a measure of quality of health care: advantages and limitations","volume":"160","author":"Benbassat","year":"2000","journal-title":"Arch Intern Med"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"912","DOI":"10.5811\/westjem.2018.8.38225","article-title":"Predictors of admission in adult unscheduled return visits to the emergency department","volume":"19","author":"Hayward","year":"2018","journal-title":"West J Emerg Med"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"1110","DOI":"10.1016\/j.jemermed.2012.01.062","article-title":"Unscheduled return visits with and without admission post emergency department discharge","volume":"43","author":"Hu","year":"2012","journal-title":"J Emerg Med"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"301","DOI":"10.11622\/smedj.2016104","article-title":"Characteristics of patients who made a return visit within 72 hours to the emergency department of a Singapore tertiary hospital","volume":"57","author":"Chan","year":"2016","journal-title":"Singapore Med J"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"346","DOI":"10.1093\/jamiaopen\/ooz019","article-title":"Predicting 72-hour and 9-day return to the emergency department using machine learning","volume":"2","author":"Hong","year":"2019","journal-title":"JAMIA Open"},{"key":"ref7","doi-asserted-by":"publisher","first-page":"86","DOI":"10.7326\/M20-3486","article-title":"Evaluation of hospital performance using the excess days in acute care measure in the hospital readmissions reduction program","volume":"174","author":"Wadhera","year":"2020","journal-title":"Ann Intern Med"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"1239","DOI":"10.1111\/j.1553-2712.2011.01223.x","article-title":"Emergency medicine-quality indicators: the United Kingdom perspective","volume":"18","author":"Heyworth","year":"2011","journal-title":"Acad Emerg Med"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"136","DOI":"10.1016\/j.annemergmed.2013.01.017","article-title":"Factors associated with short-term bounce-back admissions after emergency department discharge","volume":"62","author":"Gabayan","year":"2013","journal-title":"Ann Emerg Med"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"370","DOI":"10.1016\/j.jss.2020.08.021","article-title":"A Nested Mixed Effects Multicenter Model Examining the Risk Factors for Pediatric Trauma Return Visits Within 72 Hours","volume":"257","author":"Ehwerhemuepha","year":"2021","journal-title":"J Surg Res."},{"key":"ref11","doi-asserted-by":"publisher","first-page":"457","DOI":"10.3390\/medicina55080457","article-title":"Risk factors for emergency department unscheduled return visits","volume":"55","author":"Soh","year":"2019","journal-title":"Medicina"},{"key":"ref12","doi-asserted-by":"publisher","first-page":"947","DOI":"10.3934\/jimo.2018079","article-title":"Predicting 72-hour reattendance in emergency departments using discriminant analysis via mixed integer programming with electronic medical records","volume":"15","author":"Meng","year":"2019","journal-title":"J Ind Manag Optim"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"1592","DOI":"10.1162\/neco_a_01207","article-title":"On the complexity of logistic regression models","volume":"31","author":"Bulso","year":"2019","journal-title":"Neural Comput"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"20","DOI":"10.1186\/s12873-024-00939-6","article-title":"Machine learning models for predicting unscheduled return visits to an emergency department: a scoping review","volume":"24","author":"Lee","year":"2024","journal-title":"BMC Emerg Med"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"8063846","DOI":"10.1155\/2023\/8063846","article-title":"Machine learning approaches to predict patient\u2019s length of stay in emergency department","volume":"2023","author":"Shbool","year":"2023","journal-title":"Appl Comput Intell Soft Comput"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"224","DOI":"10.1016\/j.ajem.2024.02.013","article-title":"Navigating the landscape of medical triage: unveiling the potential and challenges of large language models and beyond","volume":"78","author":"Lansiaux","year":"2024","journal-title":"Am J Emerg Med"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"170","DOI":"10.1016\/j.ajem.2024.01.037","article-title":"Assessing the precision of artificial intelligence in ED triage decisions: insights from a study with ChatGPT","volume":"78","author":"Pasl\u0131","year":"2024","journal-title":"Am J Emerg Med"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2507.01080","article-title":"Development and comparative evaluation of three artificial intelligence models (NLP, LLM, JEPA) for predicting triage in emergency departments: a 7-month retrospective proof-of-concept","author":"Lansiaux","year":"2025","journal-title":"arXiv"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"420","DOI":"10.1016\/j.ajem.2017.08.049","article-title":"Predicting 72-hour emergency department revisits","volume":"36","author":"Pellerin","year":"2018","journal-title":"Am J Emerg Med"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"82","DOI":"10.3390\/diagnostics12010082","article-title":"A machine learning model for predicting unscheduled 72 h return visits to the emergency department by patients with abdominal pain","volume":"12","author":"Hsu","year":"2022","journal-title":"Diagnostics"},{"key":"ref21","doi-asserted-by":"publisher","first-page":"102","DOI":"10.1136\/qshc.2005.016618","article-title":"Unscheduled returns to the emergency department: an outcome of medical errors?","volume":"15","author":"Nu\u00f1ez","year":"2006","journal-title":"Qual Saf Health Care"},{"key":"ref22","doi-asserted-by":"publisher","first-page":"1584","DOI":"10.1016\/j.ajem.2019.158430","article-title":"Unscheduled return visits to the emergency department with ICU admission: a trigger tool for diagnostic error","volume":"38","author":"Aaronson","year":"2020","journal-title":"Am J Emerg Med"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"11","DOI":"10.1016\/j.ccc.2021.09.005","article-title":"Diagnostic error in the critically ill: a hidden epidemic?","volume":"38","author":"Bergl","year":"2022","journal-title":"Crit Care Clin"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"663","DOI":"10.1001\/jama.2016.0649","article-title":"In-hospital outcomes and costs among patients hospitalized during a return visit to the emergency department","volume":"315","author":"Sabbatini","year":"2016","journal-title":"JAMA"},{"key":"ref25","doi-asserted-by":"publisher","first-page":"171","DOI":"10.1007\/s11739-014-1154-8","article-title":"Overcrowding in emergency department: an international issue","volume":"10","author":"Di Somma","year":"2015","journal-title":"Intern Emerg Med"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"79","DOI":"10.1136\/emermed-2018-208343","article-title":"Characteristics and determinants of high-risk unscheduled return visits to the emergency department","volume":"37","author":"Hiti","year":"2020","journal-title":"Emerg Med J"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"21513","DOI":"10.1038\/s41598-021-00937-9","article-title":"Using explainable machine learning to identify patients at risk of reattendance at discharge from emergency departments","volume":"11","author":"Chmiel","year":"2021","journal-title":"Sci Rep"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1016\/j.cmpb.2017.03.022","article-title":"Predicting return visits to the emergency department for pediatric patients: applying supervised learning techniques to the Taiwan National Health Insurance Research Database","volume":"144","author":"Hu","year":"2017","journal-title":"Comput Methods Prog Biomed"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"850","DOI":"10.1111\/jpc.13561","article-title":"Risk of hospitalisation after early-revisit in the emergency department","volume":"53","author":"Cozzi","year":"2017","journal-title":"J Paediatr Child Health"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"1189939","DOI":"10.3389\/fpubh.2023.1189939","article-title":"Healthcare-associated adverse events and readmission to the emergency departments within seven days after a first consultation","volume":"11","author":"Gasperini","year":"2023","journal-title":"Front Public Health"},{"key":"ref31","first-page":"785","article-title":"XGBoost: A scalable tree boosting system","author":"Chen","year":"2016"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1023\/A:1010933404324","article-title":"Random forests","volume":"45","author":"Breiman","year":"2001","journal-title":"Mach Learn"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"6679","DOI":"10.1609\/aaai.v35i8.16826","article-title":"Tabnet: attentive interpretable tabular learning","volume":"35","author":"Arik","year":"2021","journal-title":"Proc AAAI Conf Artif Intell"},{"key":"ref34","doi-asserted-by":"publisher","first-page":"e22491","DOI":"10.2196\/22491","article-title":"Predicting unscheduled emergency department return visits among older adults: population-based retrospective study","volume":"9","author":"Chen","year":"2021","journal-title":"JMIR Med Inform"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"238","DOI":"10.1067\/mem.2002.121523","article-title":"Older adults in the emergency department: a systematic review of patterns of use, adverse outcomes, and effectiveness of interventions","volume":"39","author":"Aminzadeh","year":"2002","journal-title":"Ann Emerg Med"},{"key":"ref36","doi-asserted-by":"publisher","first-page":"249","DOI":"10.1111\/j.1553-2712.2000.tb01070.x","article-title":"Return to the emergency department among elders: patterns and predictors","volume":"7","author":"McCusker","year":"2000","journal-title":"Acad Emerg Med"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"e0230876","DOI":"10.1371\/journal.pone.0230876","article-title":"Risk of mortality and cardiopulmonary arrest in critical patients presenting to the emergency department using machine learning and natural language processing","volume":"15","author":"Fernandes","year":"2020","journal-title":"PLoS One"},{"key":"ref38","doi-asserted-by":"publisher","first-page":"579","DOI":"10.1097\/PEC.0b013e31828e62f4","article-title":"Unscheduled return visits to the emergency department: the impact of language","volume":"29","author":"Gallagher","year":"2013","journal-title":"Pediatr Emerg Care"},{"key":"ref39","doi-asserted-by":"publisher","first-page":"288","DOI":"10.1186\/s40001-023-01256-7","article-title":"Predictors of 72-h unscheduled return visits with admission in patients presenting to the emergency department with abdominal pain","volume":"28","author":"Lin","year":"2023","journal-title":"Eur J Med Res"},{"key":"ref40","doi-asserted-by":"publisher","first-page":"864","DOI":"10.1111\/acem.12442","article-title":"Patient returns to the emergency department: the time-to-return curve","volume":"21","author":"Rising","year":"2014","journal-title":"Acad Emerg Med"},{"key":"ref41","doi-asserted-by":"publisher","first-page":"365","DOI":"10.1111\/acem.12347","article-title":"Recurrent and high-frequency use of the emergency department by pediatric patients","volume":"21","author":"Alpern","year":"2014","journal-title":"Acad Emerg Med"}],"container-title":["Frontiers in Public Health"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2025.1609206\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T06:25:50Z","timestamp":1766125550000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2025.1609206\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,19]]},"references-count":41,"alternative-id":["10.3389\/fpubh.2025.1609206"],"URL":"https:\/\/doi.org\/10.3389\/fpubh.2025.1609206","relation":{},"ISSN":["2296-2565"],"issn-type":[{"value":"2296-2565","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,19]]},"article-number":"1609206"}}