Early Diagnosis of Poly Cystic Ovary Syndrome (PCOS) in young women: A Machine Learning Approach

Abstract

Poly Cystic Ovary Syndrome (PCOS) is the most common hormonal disorder which affects large percentage of women's in their reproductive age and also leads to serious health issues. The main symptoms include infertility, uneven menstrual cycle, mood swing, increased level of male hormones, stomach bloating, thinning of hair etc., Statistics says one in five Indian women were diagnosed with PCOS. If it is not monitored in time may cause serious health impacts. The actual cause for the PCOS is uncertain. By considering the complexities of diagnosing PCOS and time and cost incurred for diagnosis, this research article proposes an automated model which can assist for the physicians. Now a days Machine learning models are playing vital role for medical diagnosis and assisting physicians. In this paper we propose an automated model to classify PCOS and Non-PCOS women with the help of machine learning algorithms. The follicular fluid samples of 100 women was taken. And with the help of Raman spectra and efficient feature selection methods the taken data set is preprocessed. Furthermore, the performance of advanced ML classifiers like Random Forest, Ada Boost, Multilayer Perceptron and decision tree are analyzed. The implementation results reveal that Raman spectroscopy with advanced ML algorithms model can predict the PCOS with 100% accuracy with follicular fluid samples.