Advanced machine learning for estimating vascular occlusion percentage in patients with ischemic heart disease and periodontitis

Abstract

Objective

The study aimed to assess the efficacy of advanced machine learning algorithms in estimating the percentage of vascular occlusion in ischemic heart disease (IHD) cases with periodontitis.

Methods

This study involved 300 IHD patients aged 45 to 65 with stage III periodontitis undergoing coronary angiograms. Dental and periodontal examinations assessed various factors. Coronary angiograms categorized patients into three groups based on artery stenosis. Clinical data were processed, outliers were identified, and machine learning algorithms were applied for analysis using the orange tool, including confusion matrices and receiver operating characteristic (ROC) curves for assessment.

Results

The results showed that Random Forest, Naïve Bayes, and Neural Networks were 97 %, 84 %, and 92 % accurate, respectively. Random Forest did exceptionally well in identifying the severity of conditions, with 95.70 % accuracy for mild cases, 84.80 % for moderate cases, and a perfect 100.00 % for severe cases.

Conclusions

The current study, using Periodontal Inflammatory Surface Area (PISA) scores, revealed that the Random Forest model accurately predicted the percentage of vascular occlusion.