Time and event-specific deep learning for personalized risk assessment after cardiac perfusion imaging

Abstract

Standard clinical interpretation of myocardial perfusion imaging (MPI) has proven prognostic value for predicting major adverse cardiovascular events (MACE). However, personalizing predictions to a specific event type and time interval is more challenging. We demonstrate an explainable deep learning model that predicts the time-specific risk separately for all-cause death, acute coronary syndrome (ACS), and revascularization directly from MPI and 15 clinical features. We train and test the model internally using 10-fold hold-out cross-validation (n = 20,418) and externally validate it in three separate sites (n = 13,988) with MACE follow-ups for a median of 3.1 years (interquartile range [IQR]: 1.6, 3.6). We evaluate the model using the cumulative dynamic area under receiver operating curve (cAUC). The best model performance in the external cohort is observed for short-term prediction – in the first six months after the scan, mean cAUC for ACS and all-cause death reaches 0.76 (95% confidence interval [CI]: 0.75, 0.77) and 0.78 (95% CI: 0.78, 0.79), respectively. The model outperforms conventional perfusion abnormality measures at all time points for the prediction of death in both internal and external validations, with improvement increasing gradually over time. Individualized patient explanations are visualized using waterfall plots, which highlight the contribution degree and direction for each feature. This approach allows the derivation of individual event probability as a function of time as well as patient- and event-specific risk explanations that may help draw attention to modifiable risk factors. Such a method could help present post-scan risk assessments to the patient and foster shared decision-making.

Hybridizing machine learning in survival analysis of cardiac PET/CT imaging

“…Therefore, this study harnessed and engaged the advantages of traditional survival modeling through Cox regression such as interpretability (explainability), time-to-event consideration, and robustness of its well-known estimates (i.e., hazard ratios), as well as their familiarity to both clinicians and clinical researchers, while also engaging the advantage of multiple variable integration through ML analytics. Recently, this issue has also been nicely investigated by Pieszko et al 12 suggesting the inherent need for better prognostic estimates by harnessing the capacities of ML. Here, our approach was selected after considering that full ML survival modeling requires larger amounts of data and that traditional Cox regressions may not easily handle a large number of predictors in the long term (e.g., all the imaging variables) for the given amount of events recorded.…”

“…Notably, the utility of ML in predicting clinical outcomes such as mortality, revascularization, heart failure and myocardial infarction (MI) has been recently explored in both CCTA 6 and PET 11 data independently. However, there is a paucity of imaging studies regarding the implementation of ML for prognostic analysis of cardiovascular outcomes considering the influence of individual time-to-event 12 , 13 (e.g., through survival modeling through Cox proportional hazards).…”