Predicting cardiovascular events using machine learning

• Grant number: 21/06137-4
• Support Opportunities: Regular Research Grants
• Start date: April 01, 2022
• End date: March 31, 2026
• Field of knowledge: Engineering - Biomedical Engineering - Medical Engineering
• Agreement: European Commission (Horizon 2020)
• Principal Investigator: Paulo Mazzoncini de Azevedo Marques
• Grantee: Paulo Mazzoncini de Azevedo Marques
• Principal researcher abroad: Peter Rainer
• Institution abroad: University of Graz, Austria
• Host Institution: Faculdade de Medicina de Ribeirão Preto (FMRP). Universidade de São Paulo (USP). Ribeirão Preto , SP, Brazil
• Associated researchers: André Schmidt ; Elen Almeida Romão ; João Mazzoncini de Azevedo Marques ; José Abrão Cardeal da Costa ; Sandro Scarpelini
• Associated scholarship(s): 25/04100-7 - Federated learning for optimizing machine learning models trained with data from different hospital information systems., BP.PD; 24/01279-3 - Integration of Machine Learning Models into Daily Hospital Routine, BP.TT; 23/01695-4 - Validation and improvement of machine learning models for predicting cardiovascular events, BP.TT; 22/16683-9 - Federated learning for validating machine learning models trained in different hospital networks, BP.PD

Abstract

Cardiovascular disease is the leading cause of death worldwide. Underlying atherosclerosis and ensuing conditions such as myocardial infarction, ischemic heart disease and stroke cause tremendous morbidity, mortality, and economic loss. Early identification of patients with high risk for such clinical events enables preventive actions. The use of machine learning (ML) for risk prediction can outperform traditional risk scores. Although many ML models have been developed over the last years, validation is rare. We do not know how models perform in different clinical settings or populations. Furthermore, using numerous predictors, it is hard to transfer models to other health systems. Recently, we developed risk prediction models for major adverse cardiovascular events and progression of kidney disease. However, the models still lack external validation, hindering implementation in different clinical contexts and limiting generalizability. As such, this project has three main aims. The first is to validate and improve our ML models across different hospital networks and populations. The second aim is to integrate the ML models in different hospital information systems and evaluate their impact on daily hospital routine. Finally, building upon these validated models, the third aim addresses effective risk communication strategies in order to effect behavioral changes in patients. Therefore, our project makes a fundamental contribution towards employing innovative personalized risk prediction and medicine and assess its clinical implementation in a transnational context. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

SHIMIZU, GILSON YUUJI; ROMAO, ELEN ALMEIDA; CARDEAL DA COSTA, JOSE ABRAO; MAZZONCINI DE AZEVEDO-MARQUES, JOAO; SCARPELINI, SANDRO; FIRMINO SUZUKI, KATIA MITIKO; CESAR, HILTON VICENTE; AZEVEDO-MARQUES, PAULO M.. External validation and interpretability of machine learning-based risk prediction for major adverse cardiovascular events. 2024 IEEE 37TH INTERNATIONAL SYMPOSIUM ON COMPUTER-BASED MEDICAL SYSTEMS, CBMS 2024, v. N/A, p. 6-pg., 2024-01-01. (14/50889-7, 21/06137-4, 22/16683-9, 23/01695-4)

SHIMIZU, GILSON YUUJI; SCHREMPF, MICHAEL; ROMAO, ELEN ALMEIDA; JAUK, STEFANIE; KRAMER, DIETHER; RAINER, PETER P.; CARDEAL DA COSTA, JOSE ABRAO; DE AZEVEDO-MARQUES, JOAO MAZZONCINI; SCARPELINI, SANDRO; SUZUKI, KATIA MITIKO FIRMINO; et al. Machine learning-based risk prediction for major adverse cardiovascular events in a Brazilian hospital: Development, external validation, and interpretability. PLoS One, v. 19, n. 10, p. 23-pg., 2024-10-11. (21/06137-4, 22/16683-9, 23/01695-4)