Keywords
SHAP
Sepsis-3
MIMIC-IV
Dynamic landmark forecasting
How to Cite
Abstract
Sepsis-3 onset forecasting remains challenging because event prevalence is low at each patient-time landmark and retrospective electronic health record (EHR) models may inadvertently exploit workflow signals that reflect clinician suspicion rather than underlying patient physiology. We present an explainable, leakage-audited, multi-horizon LightGBM framework for physiology-based Sepsis-3 onset forecasting using MIMIC-IV v3.1. Adult first ICU stays with at least 8 hours of observation were transformed into hourly physiologic trajectories, Sepsis-3 onset labels, and repeated landmark prediction examples. Separate models estimated the probability of Sepsis-3 onset within the next 4, 6, 12, and 24 hours. The final cohort comprised 64,363 ICU stays, including 42,291 forecastable stays and 6,431 forecastable Sepsis-3 cases. The deployment-oriented no-admission-year CLEAN model achieved held-out AUROCs of 0.782, 0.772, 0.750, and 0.720 for the 4-, 6-, 12-, and 24-hour prediction horizons, respectively, with patient-level bootstrap 95% confidence intervals of 0.763–0.801, 0.752–0.792, 0.729–0.769, and 0.699–0.742. A paired leakage audit demonstrated that removing vasopressor exposure, measurement-age channels, and observation-mask channels resulted in AUROC changes of less than 0.01, indicating that predictive performance was primarily driven by physiologic information rather than workflow-related artifacts. Temporal validation on the latest deidentified admission-year group yielded a 6-hour AUROC of 0.762. At the 6-hour horizon, a top-2% no-year alarm policy with a 6-hour cooldown detected 17.8% of sepsis cases within 0–6 hours, 29.3% within 0–24 hours, and 40.3% before onset overall while generating 15.5 alerts per 100 patient-days. SHAP-based explainability analyses consistently identified neurologic status, temperature, urine-output dynamics, renal markers, inflammatory variables, vital signs, and lactate-related dynamics as the dominant predictors across forecasting horizons. These findings provide a robust and explainable physiology-based framework for future studies on Sepsis-3 onset forecasting in complex critical-care systems.
References
Boussina, A., S. P. Shashikumar, A. Malhotra, R. L. Owens, R. El-Kareh, et al., 2024 Impact of a deep learning sepsis prediction model on quality of care and survival. npj Digital Medicine 7: 14.
Duval, L., A. Villié, F. Zheng, G. Terraz, S. Blein, et al., 2025 Early prediction of vasopressor initiation in ICU sepsis patients using an interpretable EHR-based ML model. BMC Medical Informatics and Decision Making 25: 442.
Gao, J., Y. Lu, N. Ashrafi, I. Domingo, K. Alaei, et al., 2024a Prediction of sepsis mortality in ICU patients using machine learning methods. BMC Medical Informatics and Decision Making 24: 228.
Gao, Y., C. Wang, J. Shen, Z. Wang, Y. Liu, et al., 2024b Systematic review and network meta-analysis of machine learning algorithms in sepsis prediction. Expert Systems with Applications 245: 122982.
Hu, C., L. Li, W. Huang, T. Wu, Q. Xu, et al., 2022 Interpretable machine learning for early prediction of prognosis in sepsis: a discovery and validation study. Infectious Diseases and Therapy 11: 1117–1132.
Islam, K. R., J. Prithula, J. Kumar, T. L. Tan, M. B. I. Reaz, et al., 2023 Machine learning-based early prediction of sepsis using electronic health records: a systematic review. Journal of Clinical Medicine 12: 5658.
Johnson, A., L. Bulgarelli, T. Pollard, B. Gow, B. Moody, et al., 2024 MIMIC-IV. Version 3.1.
Johnson, A. E. W., L. Bulgarelli, L. Shen, A. Gayles, A. Shammout, et al., 2023 MIMIC-IV, a freely accessible electronic health record dataset. Scientific Data 10: 1.
Ke, G., Q. Meng, T. Finley, T. Wang, W. Chen, et al., 2017 LightGBM: A highly efficient gradient boosting decision tree. In Advances in Neural Information Processing Systems, volume 30, pp. 3146–3154.
Liu, Z., W. Shu, T. Li, X. Zhang, and W. Chong, 2025 Interpretable machine learning for predicting sepsis risk in emergency triage patients. Scientific Reports 15: 887.
Lu, X., H. Kang, D. Zhou, and Q. Li, 2022 Prediction and risk assessment of sepsis-associated encephalopathy in ICU based on interpretable machine learning. Scientific Reports 12: 22621.
Lundberg, S. M., G. Erion, H. Chen, A. DeGrave, J. M. Prutkin, et al., 2020 From local explanations to global understanding with explainable AI for trees. Nature Machine Intelligence 2: 56–67.
Meng, C., L. Trinh, N. Xu, J. Enouen, and Y. Liu, 2022 Interpretability and fairness evaluation of deep learning models on MIMIC-IV dataset. Scientific Reports 12: 7166.
Seymour, C. W., V. X. Liu, T. J. Iwashyna, F. M. Brunkhorst, T. D. Rea, et al., 2016 Assessment of clinical criteria for sepsis: For the third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 315: 762–774.
Shan, W., D. Sun, and Z.-P. Liu, 2024 Predicting sepsis onset in ICU patients using machine learning and feature section: a case study of MIMIC-IV data. In 2024 IEEE International Conference on Medical Artificial Intelligence (MedAI), pp. 546–551.
Shashikumar, S. P., S. Mohammadi, R. Krishnamoorthy, A. Patel, G. Wardi, et al., 2025 Development and prospective implementation of a large language model based system for early sepsis prediction. npj Digital Medicine 8: 290.
Singer, M., C. S. Deutschman, C. W. Seymour, M. Shankar-Hari, D. Annane, et al., 2016 The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 315: 801–810.
Yang, Y., B. Yi, and Y. Chen, 2026 Lightweight large language models for early sepsis prediction via a semantic abstraction rule engine. Scientific Reports.
Zhang, S.-Z., H.-Y. Ding, Y.-M. Shen, B. Shao, Y.-Y. Gu, et al., 2025 Harness machine learning for multiple prognoses prediction in sepsis patients: evidence from the MIMIC-IV database. BMC Medical Informatics and Decision Making 25: 152.
Zhang, Y., T. Kirchler, and A. P. Wang, 2026 Evaluating artificial intelligence for sepsis prediction in emergency departments: a systematic review and meta analysis. Journal of Medical Systems 50: 49.
Zubair, M., I. Din, N. Sarwar, B. Elov, and Z. Trabelsi, 2025 Revolutionizing sepsis diagnosis using machine learning and deep learning models: a systematic literature review. BMC Infectious Diseases 25: 1396.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.