Advanced Neural Decision Tree Paradigm for Proactive Detection and Precision Prediction of Polycystic Ovary Syndrome
Polycystic Ovary Syndrome (PCOS) is a prevalent endocrine system issue affecting women's fertility and health, causing significant health problems. So early detection of PCOS can help in the healing process. Early detection and prediction cannot be accomplished with the present methods and therapies. To discourse this problem, the approach suggested in this paper can aid in the early identification and forecast of PCOS care using traditional or ideal features. For pre-processing the data, along with the data cleaning, and data transformation, Tukey’s Fences-based Min-Max normalization is introduced. To diagnose PCOS, this research examines the benefits of using neural network learning to decision trees and feature selection approaches. First, the best traits that potentially predict PCOS are identified using chi-square and f-test algorithms. This feature selection methodology, which yields the greatest significant subset of structures, can rapidity up calculation and improve classifier performance. The experimental results showed that feature selection strategies improved the performance of all classifiers in a positive way by reducing the number of false negatives. Additionally, the proposed Neural decision Tree (NDT) classifier outperformed previous classifiers study and additional studies in literature with an accuracy of 97.67% by employing the compact subsection of structures based on chi-square and f-test feature selection approach.