Weight Assessment of Cardiac Risk Factors Using an Efficient Integration of FMFs and ANNs Optimization Technique

This paper presents a weight optimization of cardiac risk factors using effective integration of fuzzy membership functions (FMFs) and artificial neural networks (ANNs). The effect of long-standing high blood pressure, body mass index (BMI), low blood oxygen saturation level (Spo2), high blood cholesterol level, high Stress, smoking, alcohol, heart rate, age, diabetes or its transition on incident cardiovascular diseases (CVDs) is unclear. The objective of this study is to optimize and evaluate the weights of the cardiac risk factors associated with the development of CVDs. This study includes the opinions of 70 doctors regarding 16 cardiac affecting risk factors (CARFs). Each doctor assigns weights of CARFs based on their medical expertise in cardiology. The variability in doctors’ opinions on assigning weights to the same cardiac affecting risk factors create a confusion in selecting the optimize cardiac disease risk factor weights values. To address this challenges, useful combination of FMFs and ANNs optimization technique are employed for identifying optimized values of CARFs weights. This integrated optimization model achieved an average training loss of 3.7% and an average validation loss of 5% during fuzzy membership and ANNs weight optimization for CARFs. The most important risk factors significantly affecting cardiac health are evaluated. These include very high blood cholesterol levels (190 mg/dL and above) [0.088w], high stress levels [0.087w], obesity (BMI = 30-40 kg/m²) [0.093w], and old age (46-70 years) [0.086w]. The minimal cardiac effecting risk factors are evaluated for prehypertension [0.026w]. Artificial intelligence based techniques have developed a clinical model that predicts patients at risk of cardiovascular disease. Its straightforward methodology aids early detection of the disease and risk factors, making it highly beneficial in the clinical setting.