A Framework for Explainable Machine Learning-Based Early Disease Prediction in Healthcare

Abstract In preventive medicine, early and accurate prediction of diseases is of great importance, as it enables timely intervention, reduces mortality, improves the utilisation of clinical resources and decreases healthcare costs. However, the application of advanced machine-learning models in clinical practice is still limited due to their black-box nature, which produces highly accurate predictions but without clear and understandable rationale. In this work, we introduce a transparent clinical decision support framework that combines high-performance gradient boosting models with post-hoc explainability techniques. The proposed frame-work has been tested on four standard medical datasets, which are: UCI Heart Disease, Pima Indians Diabetes, Chronic Kidney Disease, and Wisconsin Breast Cancer. Eight machine learning algorithms were trained and validated after pre-processing (including outlier removal using the Interquartile Range method and Min-Max normalisation). For explain-ability, we used a two-level approach with global SHAP feature attribution and local LIME surrogate explanations. We found that the advanced gradient boosting models consistently beat traditional linear classifiers and individual decision trees. The best performance was obtained by CatBoost, with 91.80 % accuracy for heart disease prediction, while CatBoost and XGBoost both reached 99.17 % accuracy for chronic kidney disease classification with an excellent ROC-AUC of 0.999. Statistical validation of these improvements was done by ANOVA, Friedman and Wilcoxon signed-rank tests and found to be significant. In general, the framework provides a good trade-off between predictive accuracy and clinical interpretability and can be used as a reliable and explainable tool for healthcare decision making.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11740794/v1
Primary Topic
Artificial Intelligence in Healthcare
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

A Framework for Explainable Machine Learning-Based Early Disease Prediction in Healthcare

Md Sharfuddin, Md. Shakil
1 links
Research Square
Artificial Intelligence in Healthcare
preprint

A Framework for Explainable Machine Learning-Based Early Disease Prediction in Healthcare

Md Sharfuddin, Md. Shakil
preprint en

Abstract

Abstract In preventive medicine, early and accurate prediction of diseases is of great importance, as it enables timely intervention, reduces mortality, improves the utilisation of clinical resources and decreases healthcare costs. However, the application of advanced machine-learning models in clinical practice is still limited due to their black-box nature, which produces highly accurate predictions but without clear and understandable rationale. In this work, we introduce a transparent clinical decision support framework that combines high-performance gradient boosting models with post-hoc explainability techniques. The proposed frame-work has been tested on four standard medical datasets, which are: UCI Heart Disease, Pima Indians Diabetes, Chronic Kidney Disease, and Wisconsin Breast Cancer. Eight machine learning algorithms were trained and validated after pre-processing (including outlier removal using the Interquartile Range method and Min-Max normalisation). For explain-ability, we used a two-level approach with global SHAP feature attribution and local LIME surrogate explanations. We found that the advanced gradient boosting models consistently beat traditional linear classifiers and individual decision trees. The best performance was obtained by CatBoost, with 91.80 % accuracy for heart disease prediction, while CatBoost and XGBoost both reached 99.17 % accuracy for chronic kidney disease classification with an excellent ROC-AUC of 0.999. Statistical validation of these improvements was done by ANOVA, Friedman and Wilcoxon signed-rank tests and found to be significant. In general, the framework provides a good trade-off between predictive accuracy and clinical interpretability and can be used as a reliable and explainable tool for healthcare decision making.

Research Square
Southeast University (CN)
Good health and well-being
Artificial Intelligence in Healthcare
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon