Tri-fuzzy interval arithmetic with deep learning and hybrid statistical approach for analysis and prognosis of cardiovascular disease

In the era of artificial intelligence, healthcare informatics holds significant promise for cardiovascular disease (CVD) analysis. This study employs three computational intelligence approaches to address CVD-related challenges comprehensively. At first, various statistical methods unveil relationships between heterogeneous risk factors and predicted outcomes, employing tests of significance to discern differences in risk factors between classes with and without CVD. In the second stage, a hybrid statistical approach incorporates feature selection, identifying critical risk factors, and employs Tri-fuzzy interval arithmetic for precise estimation. Finally, the proposed Gaussian Probabilistic Neural Network (Gaussian-PNN) predicts heart disease onset with maximum accuracy, providing a nuanced assessment of CVD probability for each patient using interval-based lower and upper bounds derived from Tri-Fuzzy numbers. Experimental validations affirm the efficacy of these contributions, highlighting the analysis of significant risk factors, interrelationship establishment, and the novel integration of crisp and fuzzy interval estimates, advancing heart disease diagnosis.