نوع مقاله : اصیل پژوهشی
نویسندگان
1 گروه فیزیولوژی ورزشی، دانشکده تربیت بدنی، دانشگاه مازندران
2 گروه فیزیولوژی ورزشی، دانشکده علوم ورزشی، دانشگاه مازندران، بابلسر، ایران
3 واحد پزشکی تهران، دانشگاه آزاد اسلامی تهران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Background: Childbearing and rejuvenating society are serious challenges in today's world. Polycystic ovary syndrome (PCOS), which is associated with obesity, has become a common concern for fertility. Early diagnosis of PCOS can improve the treatment process and the quality of life of the family. The primary goal of the present study is to train machine learning models with hormonal marker data and metabolic syndrome to identify PCOS and then provide training strategies in infertile women.
Method: Data from 1000 women aged 20-45 with and without PCOS underwent initial processing, then a dataset of 500 women was prepared and important features were selected using different methods such as random forest (RF), recursive feature elimination (RFE), and mutual information (MI) and entered into different algorithms.
Results: Training the DT model with feature selection by RF method had the highest accuracy in diagnosis compared to others. Also, according to the results of feature selection by RFE, MI and RF methods, it was shown that in addition to sex hormones, fasting glucose, cholesterol, high-density lipoprotein, vitamin D3 and thyroid-related hormones are prominent factors that play a role in predicting PCOS with high accuracy.
Conclusion: Our research study will help the medical community to diagnose and treat PCOS early and provide the capacity for exercise science researchers to normalize hormonal and metabolic biomarkers with novel training approaches.
کلیدواژه‌ها [English]
)