Datasets corresponding to CHD, cancer, retinopathy, type II diabetes, and Pima Indian diabetic dataset have been used for experimentation. The observed data have been pre-processed according to the stages in data cleaning, transformation, and preparation. The process of pheromone value initialization depends on the intensity of each of the pheromone value during iteration. The relevance matrix is generated based upon the number of features used for the dataset during experimentation. The relevance and the similarity matrix play a significant role upon initialization of a number of features in the observed medical data. The generation of population with ACO depends upon the number of features that has to be fed as input for the algorithmic model. The heuristic information gets updated in accordance with the search process by the ants with pheromone values. The algorithmic model must have the ability to determine the minimum number of relevant and dependent features with improved accuracy.