Enhanced Polycystic Ovary Syndrome diagnosis model leveraging a K-means based genetic algorithm and ensemble approach

Polycystic Ovary Syndrome (PCOS) is a prevalent hormonal disorder affecting women in their childbearing years. Detecting PCOS early is crucial for preserving fertility in young women and preventing long-term health complications like hypertension, heart disease, and obesity. While costly clinical tests exist to detect PCOS, there is a growing demand for more accurate and affordable diagnostic methods. The primary objective of this research is to pinpoint the most effective PCOS features that can aid experts in early diagnosis. We introduce a feature extraction model, termed KM-GN, which combines the k-means algorithm with a genetic selection algorithm to identify the most informative features for PCOS detection. These selected features are fed into our designed model, Random Subspace-based Bootstrap Aggregating Ensembles (RSBE). To assess the performance of the proposed RSBE method, we compare it against several individual and ensemble classifiers. The effectiveness of our model is assessed using a freely accessible dataset comprising 43 traits from 541 women, of whom 177 have been diagnosed with PCOS. We employ various statistical metrics to evaluate the performance, including the confusion matrix, accuracy, recall, F1 score, precision, and specificity. The experimental outcomes demonstrate the viability of implementing our proposed model as a hardware tool for efficient detection of PCOS.