Stacking Ensemble Machine Learning for Cardiac Safety Assessment Using hiPSC-CM MEA Data

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Perfect discrimination: AUC = 1.000 for stacking ensemble

• Primary Outcome: TdP risk prediction accuracy using stacking ensemble vs single classifiers

• Nnt Or Sensitivity: Pairwise accuracy = 1.000, classification error <0.125

• Confidence Interval: Not reported

Clinical Application

Requires specialised MEA equipment and hiPSC-CM culture facilities, limiting accessibility to research institutions Relevant to TGA drug approval processes and pharmaceutical R&D in Australia. Could support ICH S7B/E14 cardiac safety requirements Pharmaceutical companies conducting early-stage cardiac safety screening of drug candidates

Abstract

PURPOSE: Prediction of Torsades de Pointes (TdP) risk using hiPSC-CM assays remains challenging, as many models fail to capture nonlinear patterns and exhibit unstable performance across different drugs. We examined whether a stacking ensemble can improve robustness when only two simple Δ F P D c -derived predictors are available. METHODS: Two electrophysiological predictors were derived from MEA-based FPDc measurements: the maximum change ( Δ F P D c ) and the interpolated Δ F P D c at Cmax. These features were used to train a stacking model comprising random forest (RF), XGB, and a shallow artificial neural network (ANN). Performance was assessed on 16 unseen CiPA reference compounds using AUC, likelihood ratios, pairwise accuracy, and classification error. Class weighting addressed class imbalance, and likelihood-based metrics assessed diagnostic consistency. RESULTS: The stacking ensemble consistently outperformed all single classifiers. The XGB-based meta-classifier achieved perfect discrimination for both AUC1 and AUC2 (1.000), with pairwise accuracy reaching 1.000 and classification error remaining below 0.125 across repeated evaluations. CONCLUSIONS: Combining a simple MEA-derived feature set with a stacking architecture provides a more reliable framework for early TdP risk assessment than single-model approaches. External validation using additional MEA datasets and the integration of interpretable modeling strategies will be important for future translational use.