aceMachine learning approaches to predict early cardiac immune-related adverse events in patients receiving immune checkpoint inhibitors
Purpose Immune checkpoint inhibitor (ICI)-induced cardiac immune-related adverse events (cardiac irAEs) are rare yet serious complications. Clinical assessment tools to identify at-risk patients would allow for more effective prevention strategies, thus improving clinical outcomes. We constructed various machine learning (ML) models to predict these events among patients receiving ICI therapy. Methods A cohort of patients receiving ICI therapy from 2010 to 2023 was identified from the TriNetX database. Cardiac i…
Estabilizador Octopus para cirugia cardiaca by Alexperf. CC BY 3.0 · https://creativecommons.org/licenses/by/3.0
Researchers built and tested machine learning models to predict cardiac immune-related adverse events shortly after starting immune checkpoint inhibitor therapy. Using records for 61,117 patients from TriNetX from 2010 to 2023, they defined events by diagnosis codes within 90 days plus hospital visits and compared elastic net logistic regression, gradient boosted trees, and random forest models.
By the July 2026 publication date, the work was presented as a preliminary demonstration that risk scores could separate patients into tiers with different observed event rates, with testing AUC around 0.71-0.72. The approach suggests a path toward targeted prevention, but remains retrospective and code-based, requiring prospective validation and richer longitudinal inputs before clinical use.
- Cohort of 61,117 patients receiving ICI therapy from 2010 to 2023 identified from TriNetX database, with nearly 2% experiencing cardiac irAEs defined by diagnosis codes within 90 days of initiation with hospital visits.
- Models tested included elastic net logistic regression, gradient boosted trees, and random forest, with comparable testing performance and distinct features emphasized per SHAP values.
- Assigned risk scores stratified patients into low, medium, and high-risk tiers, with high-risk tier showing significantly higher cardiac irAE rates.
Machine learning models trained on TriNetX data stratified patients receiving immune checkpoint inhibitors into risk tiers for cardiac immune-related adverse events within 90 days, achieving moderate discrimination.
The rundown
The study used the TriNetX database to assemble patients treated with immune checkpoint inhibitors between 2010 and 2023. Cardiac irAEs were operationalized as relevant diagnosis codes occurring within 90 days of ICI initiation with corresponding hospital visits, affecting nearly 2% of the cohort.
Three modeling approaches were compared: elastic net logistic regression and tree-based methods including gradient boosted trees and random forest. Testing performance was similar across methods, and feature importance and SHAP values showed each model relied on distinct features. Risk scores were used to create low, medium, and high-risk tiers.
Sources
- Peer-reviewedSupportive Care in Cancer2026-07-29
How should this claim be treated?
The debate