Polycystic Ovary Syndrome Detection using Deep Learning

Polycystic Ovary Syndrome (PCOS) is an endocrine disorder that is commonly found in women of reproductive age. The disorder is identified by the increase of androgen, oligomenorrhea, anovulation, and sometimes due to the presence of many ingrown follicles in the ovary or the cysts. This endocrine disorder does not allow the follicle to grow for ovulation. Anovulation in women leads to an increase in weight, hair on the face and neck, missed period cycles, and can also cause high blood pressure and diabetes in the coming years. An early diagnosis of PCOS will help in reducing the chances of in fertilisation, diabetes and heart disease in women and hence we have a proposed a methodology in this paper that will help in the early diagnosis of PC OS. We have used a dataset of 541 patients that was available on Kaggle. Then the deep learning model is trained with the dataset obtained from Kaggle and the comparison of the result was done with the DT, SVM and RF. The accuracy obtained by the deep learning classifier was 98.07% and when it was compared with other functional classifiers such as SVM, DT AND RF (accuracy lies between 83-96
^%
) it was found that deep learning classifier achieved the highest accuracy. Hence, we can say that the proposed model provides an effective way of PCOS prediction. The result of the study confirms that our proposed DL model obtained the best accuracy rate among all the different classifiers used by different authors to predict PCOS. The accuracy of our proposed system can further be increased by including the omics data.