AI-ASSISTED PREDICTION OF ICU MORTALITY USING ROUTINE CLINICAL PARAMETERS: A SYSTEMATIC REVIEW
Main Article Content
Keywords
Artificial intelligence; Machine learning; Intensive care unit; Mortality prediction; Critical care; Routine clinical parameters; Explainable artificial intelligence.
Abstract
Background: Accurate mortality prediction is critical in intensive care units (ICUs) for risk stratification, clinical decision-making, and resource allocation. Traditional prognostic scoring systems have limitations in capturing the complex and dynamic nature of critical illness. Artificial intelligence (AI) has emerged as a promising approach for improving mortality prediction using routinely collected clinical data.
Objective: To systematically evaluate the performance, clinical applicability, and limitations of AI-assisted mortality prediction models developed using routine clinical parameters in ICU populations.
Methods: A systematic review was conducted following PRISMA 2020 guidelines. Searches were performed in PubMed, Scopus, Web of Science, Embase, and Google Scholar for studies published between Feb 2020 - Nov 2022. Studies investigating AI or machine learning models for mortality prediction in adult ICU patients using routinely available clinical parameters were included.
Results: Eleven studies met the inclusion criteria and were included in the final synthesis. Commonly used algorithms included XGBoost, Gradient Boosting Machine, Random Forest, CatBoost, artificial neural networks, and long short-term memory models. Reported AUROC values ranged from 0.79 to 0.92, with ensemble learning approaches demonstrating the highest predictive performance. Age, Glasgow Coma Scale score, blood urea nitrogen, serum creatinine, heart rate, respiratory rate, and oxygen saturation were consistently identified as important predictors. AI models generally outperformed conventional prognostic scoring systems while increasingly incorporating explainability techniques to improve transparency and clinical interpretability.
Conclusion: AI-assisted mortality prediction using routine clinical parameters demonstrates strong predictive performance and significant potential for supporting critical care decision-making. Further external validation and prospective implementation studies are needed to facilitate clinical adoption.
References
2. Hu CA, Chen CM, Fang YC, et al. Using a machine learning approach to predict mortality in critically ill influenza patients. BMJ Open. 2020;10:e033898.
3. Kong G, Lin K, Hu Y. Using machine learning methods to predict in-hospital mortality of sepsis patients in the ICU. BMC Med Inform Decis Mak. 2020;20:251.
4. Baker S, Xiang W, Atkinson I. Continuous and automatic mortality risk prediction using vital signs in the ICU: A hybrid neural network approach. Sci Rep. 2020;10:21282.
5. Yu K, Zhang M, Cui T, Hauskrecht M. Monitoring ICU mortality risk with a long short-term memory recurrent neural network. Pac Symp Biocomput. 2020;25:103-114.
6. Zhang Z, Ho KM, Hong Y. Predicting 30-day mortality for MIMIC-III patients with sepsis-3: A machine learning approach using XGBoost. J Transl Med. 2020;18:462.
7. Alghatani K, Ammar N, Rezgui A, Shaban-Nejad A. Predicting ICU length of stay and mortality using patient vital signs. JMIR Med Inform. 2021;9:e21347.
8. Li F, Xin H, Zhang J, et al. Prediction model of in-hospital mortality in ICU patients with heart failure. BMJ Open. 2021;11:e044779.
9. Pang K, Li L, Ouyang W, et al. Establishment of ICU mortality risk prediction models with machine learning algorithm using MIMIC-IV database. Diagnostics (Basel). 2022;12(5):1068.
10. Safaei N, Safaei B, Seyedekrami S, et al. E-CatBoost: An efficient machine learning framework for predicting ICU mortality. PLoS One. 2022;17:e0262895.
11. Li J, Liu S, Hu Y, et al. Predicting mortality in ICU patients with heart failure using an interpretable machine learning model. J Med Internet Res. 2022;24:e38082.
12. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.
13. Schardt C, Adams MB, Owens T, Keitz S, Fontelo P. Utilization of the PICOS framework to improve searching in systematic reviews. BMC Med Inform Decis Mak. 2007;7:16.
14. Bramer WM, de Jonge GB, Rethlefsen ML, Mast F, Kleijnen J. A systematic approach to searching: An efficient and complete method to develop literature searches. J Med Libr Assoc. 2018;106(4):531-541.
15. Page MJ, Moher D, Bossuyt PM, et al. PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews. BMJ. 2021;372:n160.
16. Higgins JPT, Thomas J, Chandler J, et al., editors. Cochrane Handbook for Systematic Reviews of Interventions. 2nd ed. Chichester: Wiley; 2019.
17. Wolff RF, Moons KGM, Riley RD, et al. PROBAST: A tool to assess the risk of bias and applicability of prediction model studies. Ann Intern Med. 2019;170(1):51-58.
Article Sidebar
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
All articles published in JPTCP are licensed under Copyright Creative Commons Attribution-NonCommercial 4.0 International License.
Dr. (Major) Sourabh Kumar
Assistant Professor Intensive Care, Department of Anaesthesiology, Rajendra Institute Of Medical Sciences, Ranchi, Jharkhand, India.
Dr. Kriti Prasad
2Consultant & Senior IVF Specialist, Seeds Of Innocens, Ranchi, Jharkhand, India.
Dr. Prabha Rashmi Lakra
Senior Resident, Department of Anaesthesiology, Rajendra Institute of Medical Sciences, Ranchi.
Dr. Vashist Raj
Junior Resident, Santosh Medical College, Ghaziabad, Uttar Pradesh, India.
Dr. Vikas Kumar
Junior Resident, Department of Anaesthesiology, Rajendra Institute of Medical Sciences, Ranchi, Jharkhand, India.
Dr. Shio Priye
Professor, Department of Anaesthesiology, Rajendra Institute of Medical Sciences, Ranchi, Jharkhand, India.
Dr. Ladhu Lakra
Professor & Head of Department,Department of Anaesthesiology, Rajendra Institute of Medical Sciences, Ranchi, Jharkhand, India.
