What is the clinical evidence from this study?

Study design: Cohort. Population: Five-year female breast cancer survivors (n=5131). Intervention: Deep learning survival model (DeepSurv) vs. Cox proportional hazards (CPH) model. Primary outcome: Time-dependent concordance (Ctd) index for predicting 10-year major adverse cardiovascular events (95% CI 0.712-0.779).

June 17, 2025Open Access

Deep Learning-Based Risk Prediction Model for Major Adverse Cardiovascular Events in Long-Term Breast Cancer Survivors

SOSi Nae OhYonsei University JSJae‐Yong ShimYonsei University YLYong Je LeeSeverance Hospital

Key Result

A deep learning survival model incorporating conventional and treatment-related risk factors predicted 10-year major adverse cardiovascular events in breast cancer survivors with a Ctd index of 0.738.

Study Design

Type

Cohort (n=5,131)

Structured PICO

Does a deep learning survival model improve the prediction of 10-year MACEs compared to a traditional Cox proportional hazards model in five-year breast cancer survivors?

Population

5,131 female breast cancer survivors who survived at least 5 years post-diagnosis, followed for up to 10 years to predict major adverse cardiovascular events.

Exposure

Deep learning survival model (DeepSurv)

Comparator

Traditional Cox proportional hazards (CPH) model

Outcome

Occurrence of major adverse cardiovascular events (MACEs) at any time before the final follow-up at 10 yearscomposite

A deep learning survival model incorporating conventional and breast cancer treatment-related risk factors outperformed a traditional Cox model in predicting 10-year MACEs in five-year breast cancer survivors.

Main Result

Absolute Event Rate: 0.738% vs 0.733%

Limitations

Low use of HER-2 targeted therapy (0.7%) limits generalizability to contemporary breast cancer populations.
Reliance on administrative data precluded inclusion of factors like radiation laterality, family history of CVD, echocardiographic parameters, ECG, biomarkers, and genetics.
Requires prospective external validation before clinical implementation.

Abstract

Abstract Background Clinical practice guidelines recommend cardiovascular toxicity risk restratification including evaluation of new cardiovascular risk factors and cardiovascular diseases 5 years after cancer therapy in adult cancer survivors. We aimed to develop prediction models for the risk of 10-year cardiovascular complications in five-year breast cancer survivors. Methods We used the Korean National Health Insurance Service databases between 2005 and 2021, including 5,131 five-year female breast cancer survivors diagnosed in 2006. The study population was randomly split in a 4:1 ratio into the derivation and validation cohort. The primary outcome was the occurrence of major adverse cardiovascular events (MACEs) at any time before the final follow-up at 10 years. We developed a deep learning survival model (DeepSurv) and compared its performance with traditional Cox proportional hazards (CPH) model using the same dataset. Model performance was assessed by time-dependent concordance (C td ) index. Shapley Additive Explanations (SHAP) was used to assess feature importance. Results In the validation cohort, the DeepSurv and CPH model yielded C td index values of 0.738 (95% CI, 0.712–0.779) and 0.733 (95% CI, 0.649–0.777), respectively. In SHAP analysis, age, history of stroke, dyslipidemia, anthracycline use, and aromatase inhibitor therapy ranked highly in both models. Conclusions A deep learning survival model that incorporates both conventional and breast cancer treatment-related cardiovascular risk factors outperformed traditional regression model in predicting 10-year MACEs among individual five-year breast cancer survivors.

Ask AI

Helpful

Bookmark

View Full Paper