Machine learning–optimized discharge timing in typhoid care: Implications for clinical outcomes, cost efficiency, and health system performance

Shekoofeh Sadat Momahhed and Atefehsadat Haghighathoseini

PLOS ONE, 2026, vol. 21, issue 7, 1-49

Abstract: Objective: To develop and validate a machine learning model identifying typhoid fever patients at risk of unnecessarily prolonged hospitalization, and to quantify the clinical, financial, and systemic consequences of model-guided discharge optimization in a fragmented, multi-payer public insurance system. Setting: Iran Health Insurance Organization inpatient claims for typhoid fever, all provinces, 2024–2025; 80,223 raw claims aggregated into 13,105 hospitalization episodes. Design: Retrospective observational study with embedded computational modelling. A gradient-boosted classifier was developed on administrative claims and validated through stratified five-fold cross-validation. Predictions were translated into clinical, financial, and systemic impact levels. Results: The model achieved an area under the receiver operating characteristic curve of 0.862, outperforming logistic regression (0.831). Among 13,105 episodes, 4,658 patients were identified as potentially having reducible length of stay, projecting 6,459 potentially recoverable bed-days pending clinical review. Restricted to 5,784 inpatient admissions, optimized discharge timing reduced total expenditure by 22.9% (USD 1,306,306, purchasing power parity-adjusted); 75.0% of savings accrued to patients as reduced out-of-pocket payments and only 25.0% to the insurer, reflecting a mean insurance coverage rate of 27.7% for this care pathway. Freed capacity could accommodate 1,328 additional admissions without infrastructure expansion. Public hospital patients and those aged 60 years and above showed the greatest benefit, with out-of-pocket reductions of 43.8% and 35.5% respectively. The strongest predictor of prolonged stay was payer-classification status rather than any clinical variable, revealing a structural misalignment between insurance administration and discharge practice. Conclusion: Machine learning applied to routine insurance claims can reliably flag patients for clinical discharge review and quantify multi-dimensional system impacts. Discharge optimization in typhoid care functions primarily as a patient financial protection intervention, and should be prioritized in public hospitals and among elderly patients where financial vulnerability and optimization potential are both greatest.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0354148 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 54148&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0354148

DOI: 10.1371/journal.pone.0354148

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().