Machine learning models improve mortality prediction in patients with STEMI

Disclosures: Gupta does not report any relevant financial information. Qamar reports receiving an institutional grant from Novo Nordisk and the NorthShore Auxiliary Research Scholar Fund, and fees for educational activities from the American College of Cardiology, Clinical Exercise Physiology Association, Janssen, Medscape, Pfizer, the Society for Vascular Medicine and the Society for Angiography and Cardiovascular Interventions. Please see the study for relevant financial information from other authors.

We have not been able to process your request. Please try again later. If you continue to have this problem, please contact

Machine learning models have improved prediction of mortality after STEMI in low-income countries, researchers report in the International Journal of Cardiology.

“Several well-validated population-based risk scores such as TIMI and GRACE are available to assess mortality after STEMI. However, these risk scores are based on logistic regression analysis, which has a number of shortcomings,” Mohit D. Gupta, MD, DM (Cardiology), FRCP (Edin), FACC, FSCAI, FESC, FCSI, FAPI, Professor of Cardiology at GB Pant Institute of Post Graduate Medical Education and Research New Delhi, and Arman Qamar, MD, MPH, interventional cardiologist and adjunct researcher at NorthShore Cardiovascular Institute, NorthShore Health System, Evanston, Illinois, Healio told Healio. “Furthermore, these scores have been postulated and validated in the Caucasian population. This prompted us to use data from the NORIN-STEMI registry to assess risk factors specific to low- and middle-income countries such as India and to develop an artificial intelligence (AI) model based on it.

Heart Matrix_Adobe Stock

Source: Adobe Stock

Model tests

Using data from the NORIN-STEMI Prospective Registry of North Indian patients with STEMI, the researchers tested and trained all available machine learning models and chose five for analysis. The study population included 3191 patients (mean age, 54 years; 16% female), of whom 2668 were included in the development dataset and 523 were included in the validation dataset.

Arman Kamar

“These five were shortlisted based on the consistency of machine learning model results on an unseen dataset and AI models for which AI interpretability algorithms can be integrated,” Gupta said. and Qamar in Healio.

The goal was to make mortality predictions as accurate as possible, with high sensitivity in the validation data set.

The 30-day mortality rate was 7.7%.

The Extra Tree Classifier, a tree-based method, had the best predictive performance of the models, with a sensitivity of 85%, an accuracy of 75% and an area under the curve of 79.7%, according to the researchers.

Sensitivity “of the utmost importance”

“The main reasons for these results could be the structure of the data, its characteristics – categorical and continuous variables,” Gupta and Qamar told Healio. “In addition, [Extra Tree] the model consistently performed better in terms of sensitivity and AUC. Even though other machine learning models had good accuracy, it is the sensitivity of the model that is of paramount importance that needs to be considered for risk stratification in low- and middle-income countries. Thus, we finalized the Extra Tree Classifier as a MERC model for risk prediction after STEMI.

“Although the use of AI raises some concerns, particularly in terms of ethical issues related to data and the idea of ​​being more dependent on machines than on the human brain, our message is quite clear: AI does not will not replace the role of a doctor, but on the contrary, it will strengthen his wisdom to make a wise decision,” they said.

Using an interpretable SHaP summary graph, researchers determined that predictors of 30-day mortality included time to revascularization, cardiogenic shock, left ventricular ejection fraction <30%, age, serum creatinine, CI, female sex and moderate to severe mitral regurgitation (P < .001 for all).

“In resource-limited settings, sorting out healthcare resources is of the utmost importance,” Gupta and Qamar told Healio. “Resource use can only be appropriate when there has been adequate risk stratification by prioritizing healthcare resources to those most at risk of adverse events. This would help to efficiently and effectively utilize the limited resources available in health care. For STEMI care, these resources would include intensive care beds, prompt medical care, regular and rigorous follow-up, and the need for term monitoring devices. With the ever-increasing rate of STEMI in resource-limited countries like India, prioritization of healthcare resources is all the more necessary. This requires risk stratification to identify high-risk patients requiring priority allocation of resources.

For more information:

Mohit D. Gupta, MD, DM (Cardiology), FRCP (Edin), FACC, FSCAI, FESC, FCSI, FAPI, can be contacted at

Arman Qamar, MD, MPH, can be reached at; Twitter: @aqamarmd.


Leave a Comment