Multi-domain Identification of Myocardial Infarction Incidence using Explainable AI: The Overlooked Role of Periodontal Health
Hussein, M.; Li, L.; Falkner, K.; Buck, M.; Diaz, P.; Hatzikirou, H.
Show abstract
Myocardial infarction (MI) is a major global health concern influenced by diverse risk factors. Despite growing evidence of oral- systemic connections, current MI models largely exclude oral health indicators, reflecting the longstanding separation between dental and medical paradigms. This study introduces a multidomain, interpretable machine learning framework that integrates detailed periodontal and oral hygiene variables, marking one of the first efforts to quantitatively incorporate these features into MI incidence identification. A population-based case-control dataset comprising 1,355 individuals and heterogeneous variables was used to train and evaluate seven supervised classifiers via nested cross-validation. Among them, XGBoost achieved the best performance (AUC = 0.88{+/-}0.01; F1 score = 0.74{+/-}0.03) and was further probability-calibrated using isotonic regression, yielding a mean Brier score of 0.14{+/-}0.01 and demonstrating well-aligned predicted probabilities. SHAP values confirmed the importance of conventional cardiovascular predictors, while several periodontal indicators such as mean clinical attachment loss, plaque index, and gingival bleeding emerged among the most influential features. Sex-stratified SHAP analysis revealed sex-specific patterns in the relative impact of oral features. Additionally, individual-level waterfall plots illustrated how oral inflammation may contribute independently or in combination with conventional factors to MI incidence identification. These findings support a systems-level view of periodontitis as a modifiable, biologically relevant factor in cardiovascular health and underscore the value of considering oral-health markers within screening and management frameworks.
Matching journals
The top 2 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Identification of predictive patient characteristics for assessing the probability of COVID-19 in-hospital mortality 92%
- Uncovering the effects of model initialization on deep model generalization: A study with adult and pediatric chest X-ray images 91%
- Enhancing Fairness in Disease Prediction by Optimizing Multiple Domain Adversarial Networks 90%
Similar papers in this journal
- Selecting the most important self-assessed features for predicting conversion to Mild Cognitive Impairment with Random Forest and Permutation-based methods 94%
- Machine learning for classifying chronic kidney disease and predicting creatinine levels using at-home measurements 93%
- Mitigating Machine Learning Bias Between High Income and Low-Middle Income Countries for Enhanced Model Fairness and Generalizability 92%
Similar papers in this journal
- Enhanced machine learning and hybrid ensemble approaches for coronary heart disease prediction 93%
- A machine learning approach to identifying important features for achieving step thresholds in individuals with chronic stroke 93%
- Integrating feature importance techniques and causal inference to enhance early detection of heart disease 92%
Similar papers in this journal
Similar papers in this journal
- Using random forests to uncover the predictive power of distance-varying cell interactions in tumor microenvironments 92%
- Bayesian Structural Time Series for Biomedical Sensor Data: A Flexible Modeling Framework for Evaluating Interventions 92%
- Contrasting factors associated with COVID-19-related ICU admission and death outcomes in hospitalised patients by means of Shapley values 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.