A feature selection method utilizing path accumulation cost, redundancy minimization, and interaction maximization for the diagnosis of coronary heart disease.

Background: Coronary heart disease (CHD) is a major cause of mortality worldwide, with an increasing trend of affecting younger populations. The asymptomatic early stages and rapid progression of CHD make diagnosis challenging, necessitating efficient diagnostic approaches. Methods: We propose a novel algorithm that focuses on accumulating soft path costs to discern crucial indicators from extensive diagnostic tests, aiming to improve early CHD identification. Our approach emphasizes feature interaction using an interaction accumulation evaluation function to identify features with maximal interaction and minimal redundancy. A new stopping criterion based on information gain ratio is also introduced. Results: Experimental outcomes demonstrate that our algorithm outperforms several classical algorithms in terms of classification accuracy and feature dimension reduction, while also identifying highly correlated feature subsets. Conclusion: The proposed approach offers an efficient solution for early detection of CHD by identifying critical indicators, reducing diagnostic complexity, and improving predictive accuracy, thus potentially leading to more effective CHD management.