SHAP enhanced predictive modeling for clinical machine learning.
- 1Institute for Clinical and Translational Research, Carbone Cancer Center, School of Medicine and Public Health, University of Wisconsin-Madison, 600 Highland Avenue, Madison, WI 53792 USA.
- 2Department of Biostatistics and Data Science, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, KS 66160 USA.
- 3Hematology, Oncology and Palliative Care, Institute for Clinical and Translational Research, Carbone Cancer Center, School of Medicine and Public Health, University of Wisconsin-Madison, 600 Highland Avenue, Madison, WI 53792 USA.
Abstract
Clinical prediction is challenged by heterogeneity and noise in healthcare data. We evaluated SHAP-Enhanced Predictive Modeling (SEPM), a data-centric framework that ranks training observations using aggregated absolute Shapley Additive Explanations (SHAP) and optimizes subset retention through cross-validation. Across four clinical datasets and five classifiers, SEPM was compared with full-data training (TML) and randomized subset selection (SEPM-RAND) using nested cross-validation. A separate common-protocol XGBoost benchmark compared SEPM with Truncated Monte Carlo (TMC) Data Shapley estimation. Twenty simulation scenarios varied training sample size and outcome prevalence, with evaluation on full, high-SHAP, and low-SHAP test groups. SEPM achieved higher mean F1 scores than TML in 18 of 20 dataset-classifier combinations and SEPM-RAND in 16, retaining 42.1%-94.4% of training observations. TMC achieved higher mean F1 scores and balanced accuracy in all four datasets, whereas SEPM had lower probability-error measures and pipeline execution times (TMC-to-SEPM runtime ratios: approximately 65-1,216). The dataset-level F1 comparison was not significant (exact two-sided Wilcoxon p = 0.125). SEPM achieved the highest mean full-test F1 in all 20 simulated scenarios, but higher mean expected calibration error and Brier scores. SEPM may improve threshold-based classification with fewer training observations, but its comparison with randomized selection does not isolate ranking at identical final subset sizes, and probability quality can deteriorate. External validation, fixed-size comparisons, subgroup assessment, and calibration-aware optimization are needed before clinical use. Clinical trial registration Not applicable. The online version contains supplementary material available at https://doi.org/10.1007/s44250-026-00440-w.
Related
- A machine learning-based predictive model for gestational diabetes mellitus based on first trimester clinical indicators
- AI-driven predictive modeling of supercritical CO₂ solubility using cheetah-optimized machine learning.
- Evaluation of Perioperative Cardiovascular Event Risk in Gastrointestinal Surgery ― Predictive Modeling and Risk Stratification Using Machine Learning ―
- Predictive modeling of antenna performance using machine learning for IoT applications
- Development and internal validation of a predictive model for clinical deterioration in newly diagnosed lung cancer patients based on interpretable machine learning.
BibTeX
@article{amponsah2026shap,
title = {{SHAP enhanced predictive modeling for clinical machine learning.}},
author = {Jonah K Amponsah and Jeffrey A Thompson and Shellie D Ellis},
journal = {Discover health systems},
year = {2026},
month = nov,
doi = {10.1007/s44250-026-00440-w},
url = {https://europepmc.org/article/MED/42996766},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - SHAP enhanced predictive modeling for clinical machine learning.
AU - Jonah K Amponsah
AU - Jeffrey A Thompson
AU - Shellie D Ellis
PY - 2026
DA - 2026/11/05
JO - Discover health systems
DO - 10.1007/s44250-026-00440-w
UR - https://europepmc.org/article/MED/42996766
AB - Clinical prediction is challenged by heterogeneity and noise in healthcare data. We evaluated SHAP-Enhanced Predictive Modeling (SEPM), a data-centric framework that ranks training observations using aggregated absolute Shapley Additive Explanations (SHAP) and optimizes subset retention through cross-validation. Across four clinical datasets and five classifiers, SEPM was compared with full-data training (TML) and randomized subset selection (SEPM-RAND) using nested cross-validation. A separate common-protocol XGBoost benchmark compared SEPM with Truncated Monte Carlo (TMC) Data Shapley estimation. Twenty simulation scenarios varied training sample size and outcome prevalence, with evaluation on full, high-SHAP, and low-SHAP test groups. SEPM achieved higher mean F1 scores than TML in 18 of 20 dataset-classifier combinations and SEPM-RAND in 16, retaining 42.1%-94.4% of training observations. TMC achieved higher mean F1 scores and balanced accuracy in all four datasets, whereas SEPM had lower probability-error measures and pipeline execution times (TMC-to-SEPM runtime ratios: approximately 65-1,216). The dataset-level F1 comparison was not significant (exact two-sided Wilcoxon p = 0.125). SEPM achieved the highest mean full-test F1 in all 20 simulated scenarios, but higher mean expected calibration error and Brier scores. SEPM may improve threshold-based classification with fewer training observations, but its comparison with randomized selection does not isolate ranking at identical final subset sizes, and probability quality can deteriorate. External validation, fixed-size comparisons, subgroup assessment, and calibration-aware optimization are needed before clinical use. Clinical trial registration Not applicable. The online version contains supplementary material available at https://doi.org/10.1007/s44250-026-00440-w.
SN - 2731-7501
ER -