Cooray, R., Kaniappan Chinnathai, M. and Chahed, S. 2025. PCOS Care: A Machine Learning-Based Web Application for Early Risk Prediction of Polycystic Ovary Syndrome. 2025 IEEE International Conference on Advanced Healthcare Systems (IEEE ICAHS 2025). Hammamet, Tunisia 05 - 06 Dec 2025 IEEE . https://doi.org/10.1109/ICAHS66950.2025.11553077
| Title | PCOS Care: A Machine Learning-Based Web Application for Early Risk Prediction of Polycystic Ovary Syndrome |
|---|---|
| Authors | Cooray, R., Kaniappan Chinnathai, M. and Chahed, S. |
| Type | Conference paper |
| Abstract | Polycystic Ovary Syndrome (PCOS) is a common but underdiagnosed hormonal disorder affecting women of reproductive age, with delayed diagnosis contributing to significant long-term health complications. Although various diagnostic methods exist, there remains a lack of accessible, predictive, and user-centered tools that support its early screening and risk assessment, especially for underserved populations. To address this gap, this paper presents PCOS Care, a Machine Learning (ML) based web application designed to provide early risk prediction of PCOS through a publicly available user-friendly screening. The system utilizes two levels of risk prediction, a simple model based on lifestyle and physical indicators, and an enhanced model incorporating hormonal test data. Models were trained on real-world data, publicly available on Kaggle, using supervised learning classification algorithms such as Logistic Regression, Random Forest, Support Vector Machine (SVM) and Gradient Boosting with hyperparameter tuning and 10-fold-cross-validation. The enhanced model achieved the best performance with Random Forest, reporting 93.75% recall and 92.66% accuracy. The simple model also demonstrated strong potential, with 83.33% recall and 81.82% accuracy with Logistic Regression. Feature selection was performed using statistical methods such as analysis of variance (ANOVA), chi-square, correlation analysis, and mutual information to select the most relevant predictors. The application integrates a Streamlit-based frontend with a Flask -backend and is deployed on Render using a continuous integration and continuous deployment (CI/CD) pipeline. The usability of the system was tested with 32 potential users, and the results showed high acceptance, ease of navigation and system reliability. Conclusively, PCOS Care illustrates the benefits of ML in healthcare by enabling preliminary risk assessments and empowering users with a digital tool for informed health decisions. |
| Keywords | Polycystic Ovary Syndrome, Machine Learning, Reproductive Health, Risk Prediction, Medical Diagnosis, SMOTE, Streamlit, Flask API, Logistic Regression, Healthcare Technology |
| Year | 2025 |
| Conference | 2025 IEEE International Conference on Advanced Healthcare Systems (IEEE ICAHS 2025) |
| Publisher | IEEE |
| Accepted author manuscript | File Access Level Open (open metadata and files) |
| Publication dates | |
| Published | 03 Aug 2026 |
| ISBN | 9798331599867 |
| Digital Object Identifier (DOI) | https://doi.org/10.1109/ICAHS66950.2025.11553077 |
| Web address (URL) of conference proceedings | https://ieeexplore.ieee.org/xpl/conhome/11552943/proceeding?sortType=vol-only-seq&isnumber=11552957&pageNumber=3 |
| Web address (URL) | https://ieeexplore.ieee.org/document/11553077 |