Sep 2026· Health Science Reports· Vol 9 9, pp.
e73100
· 0 citations
Medicine
TL;DR
Combining hybrid ensemble learning, two-stage feature selection, and XAI approaches provides a computationally efficient, dependable, and interpretable method for PCOS diagnosis and practitioners may find this model to be a useful decision-support tool that improves the accuracy of diagnosis and lessens the need for human interpretation.
Abstract
Background and Aims
Polycystic ovary syndrome (PCOS) is the most prevalent endocrine ailment impacting women of reproductive age, distinguished by ongoing imbalances of hormones that lead to the growth of various ovarian cysts alongside other health issues. Infrequent or delayed menstrual periods, and also frequently excessive amounts of androgen, a male hormone, are characteristics of women with PCOS. Despite other consequences, this disease may lead to weight gain, undesirable body hair, type 2 diabetes, and gestational diabetes. The number of cases of PCOS has shockingly ascended in the past few years, according to statistics. But the clinical diagnosis procedure for PCOS in the real world becomes critical since the precision of interpretations greatly relies on the knowledge of the doctor. Therefore, a PCOS prediction model using artificial intelligence might be an appropriate supplement to the lengthy and error-prone diagnosis method.
Methods
To strictly prevent data leakage, the dataset underwent an 80/20 train-test split before applying SMOTE and two-stage feature selection exclusively to the training data. This framework combined Chi-Square filtering with Recursive Feature Elimination to identify the most pertinent features. A hybrid XGBoost (XGB) and Multi-Layer Perceptron (MLP) ensemble was selected from multiple machine learning models following rigorous evaluation using 10-fold stratified cross-validation. Finally, SHapley Additive exPlanations and Local Interpretable Model-Agnostic Explanations were integrated to significantly enhance overall model transparency.
Results
The hybrid ensemble model (XGB + MLP) outperformed conventional single ML models, with the greatest classification accuracy of 96.33% among all evaluated models. The model's interpretability and performance were greatly enhanced by the 16 characteristics that were chosen.
Conclusion
Combining hybrid ensemble learning, two-stage feature selection, and XAI approaches provides a computationally efficient, dependable, and interpretable method for PCOS diagnosis. Practitioners may find this model to be a useful decision-support tool that improves the accuracy of diagnosis and lessens the need for human interpretation.
Polycystic Ovary Syndrome (PCOS) is the most prevalent endocrine disorder affecting women of reproductive age, and there is a need for early and accurate diagnosis to prevent long-term reproductive and metabolic consequences. This work introduces a transparent ensemble model to predict PCOS at the individual level using everyday clinical, hormonal, metabolic, and lifestyle parameters. This work allows patient-based prediction, as this includes biologically plausible predictors such as serum testosterone, the luteinizing hormone to follicle-stimulating hormone ratio (LH/FSH), insulin, and menstrual irregularities. A scheme for data preprocessing, feature selection, model training, and testing is proposed. The performance of four classical classifiers, i.e., Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Gradient Boosting (GB), is evaluated. Then, a voting-based ensemble method is proposed to enhance robustness and generalization. Results of experiments confirm good predictive performance, even for a significantly challenging task, with 96% accuracy and an ROC–AUC of 0.99, while decreasing false-negative rates is highly important for early screening. To add a layer of transparency and clinical trustworthiness, SHAP-based explainable artificial intelligence was adopted to evaluate global and patient-level feature importance. In addition to binary prediction, the proposed model reduces risk stratification to a probabilistic scale (low, moderate, high), making it more practical for clinical decision support. In conclusion, our proposed explainable ensemble framework presents a scalable, accurate, and interpretable decision support system for PCOS that is feasible to adopt in practical healthcare environments, especially in rural areas where medical resources are limited and more medical aids for the detection of such diseases are desperately needed. It has strong potential for integration into AI-enabled clinical screening systems.
Aruna Pavate, Tanvi Sawant, S. Vhatkar· Turkish Journal of Engineeri...· 0 citations
Polycystic ovary syndrome (PCOS) is a very common endocrine disorder that affects women of child bearing age across the globe with a global prevalence of between 8-13 percent. Although it is a common disease, PCOS is highly undiagnosed, as the clinical picture is heterogeneous, and the diagnosis is based on costly hormonal tests and ultrasound studies. Later diagnosis causes chronic complications such as infertility, diabetes mellitus type 2, heart diseases, and endometrial cancer. This article reports a machine learning model of early and accurate prediction of PCOS based on an ensemble of Random Forest and XGBoost classifiers that will be trained on a multi-feature clinical dataset of hormonal measurements, anthropometric measurements, ultrasonographic results, and lifestyle factors. The dataset, which is obtained on the publicly accessible Kaggle PCOS dataset (n=539 patients, n=41 features), was preprocessed with a severe approach such as missing value imputation, label encoding, feature selection through Recursive Feature Elimination (RFE), and Synthetic Minority Oversampling Technique (SMOTE) to handle the imbalance in the classes. Experimental performance shows that XGBoost has the best performance with an accuracy of 92.8, F1-score of 0.922, and AUC-ROC of 0.971 whilst the random forest has an accuracy of 91.3 and AUC-ROC of 0.963 and is way better than the baseline classifiers such as logistic regression (81.2) and The most discriminative predictors based on the analysis of feature importance are follicle count, anti-Mullerian hormone (AMH) level, FSH:LH ratio, and cycle irregularity. The suggested system presents an efficient, non-invasive, and scaled clinical decision support tool in gynaecology practice.
Thee pika, S. Megala, My thili et al.· International Journal of Adv...· 0 citations
Polycystic Ovary Syndrome (PCOS) is a prevalent endocrine disorder affecting reproductive-age women, leading to infertility, hormonal imbalance, insulin resistance, and cardiovascular complications. Early diagnosis remains challenging due to heterogeneous manifestations, overlapping symptoms, and lack of automated screening tools. To address these issues, this study presents a comprehensive comparative framework for PCOS prediction using machine learning and deep learning on a public dataset of 541 patient records. The framework incorporates missing value imputation, feature standardization, SMOTE class balancing, and correlation-based feature selection. Five machine learning algorithms (Decision Tree, KNN, SVM, Random Forest, XGBoost) and two deep learning architectures (LSTM, CNN-ResNet) were evaluated using accuracy, precision, recall, F1-score, and ROC-AUC, with confusion matrices employed for detailed classification analysis. KNN achieved the highest accuracy (92.66\%), recall (91.67\%), and F1-score (89.19\%), while XGBoost delivered the best discriminative capability (ROC-AUC = 0.9540). Among deep learning models, LSTM consistently outperformed CNN-ResNet, demonstrating superior ability to capture complex clinical feature relationships. SHAP-based explainability identified ovarian follicle counts, menstrual irregularities, hair growth, weight gain, skin darkening, and anti-Müllerian hormone (AMH) levels as the most influential predictors. These findings indicate that explainable machine learning models, particularly KNN and XGBoost, provide accurate and interpretable decision support for early PCOS screening, enabling timely intervention and offering a promising foundation for intelligent healthcare decision-support systems.
Sana Rubab, Musarrat Shaheen, Zohrain Tabassum et al.· Biomedical Informatics and S...· 0 citations
Polycystic ovary syndrome (PCOS) is a prevalent endocrine disorder in women of reproductive age, yet early
risk stratification remains challenging due to its multifactorial nature. This paper proposes an AI-driven
wearable framework for early PCOS prediction built around a novel Neuroendocrine Stress Index (NSI) — a
composite biomarker aggregating pulse rate, thyroid stimulating hormone, prolactin, random blood sugar, and
BMI — which captures the cumulative physiological stress burden associated with PCOS onset. The framework
integrates clinical, hormonal, metabolic, and wearable-derived signals (heart rate, sleep duration, physical
activity, stress level) from the Kaggle PCOS dataset, processed through a structured pipeline of median
imputation, standardization, and feature selection. An ensemble of Random Forest, Logistic Regression, and
XGBoost classifiers is trained and evaluated, achieving up to 83.4% accuracy. SHAP-based explainability
identifies follicle count, AMH, BMI, and the NSI as the most influential predictors, enhancing clinical
transparency. Results confirm that embedding stress-related wearable signals alongside conventional clinical
features improves prediction accuracy and supports personalized, interpretable healthcare decision-making for
women at risk of PCOS.
K. R. Revathy, S. Padmapriya, D. Meenakshi· International Journal of Dru...· 0 citations
Polycystic Ovary Syndrome (PCOS) is a disease that plagues 8 to 13% of women of child bearing age; though only about 70 percent of the cases are diagnosed. Available machine-learning (ML) studies on PCOS detection rarely consider issues of multicollinearity in sets of hormonal features, fail to explicitly rectify the imbalance of classes and lack clinically interpretable prediction. The current research proposes a single end-to-end and reproducible ML pipeline, which simultaneously addresses all these three shortcomings. On a combined dataset (n=541 patients in Kerala, India) of merged patient data, we use iterative Variance Inflation Factor (VIF) elimination to reduce 41 raw features to 19 predictors that do not exhibit multicollinearity (all VIF< 5). Synthetic Minority Oversampling Technique (SMOTE) is then used to correct a 2:1 imbalance in training set classes. There are four variants of XGBoost, two variants of CatBoost, and K-Nearest Neighbours whose classifiers are benchmarked under the same experimental conditions on a held-out test set of 162 records. Accuracy of the SMOTE-tuned XGBoost model is 90.12%, with precision of 84 percent in PCOS-positive, recall of 84 percent, and an F1-score of 0.84, which is better than CatBoost (88.27 per cent) and KNN (86.42 per cent). SHAP (SHapley Additive exPlanations) analysis supports the assertion that the quantity of follicles, FSH/LH ratio, and the level of LH should be considered the most prevalent predictors, and that the evidence provided by the analysis can be interpreted by clinicians and corresponds to the Rotterdam diagnostic criteria. Such a pipeline provides a proven and multicollinearity-controlled foundation that supports reproducible PCOS screening studies.
Arya Malode, Vijayshri Injamuri, Vikul Pawar et al.· International journal of com...· 0 citations
Polycystic ovary syndrome (PCOS) is one of the commonest endocrine disorders in women of reproductive age, yet diagnosis is often delayed because symptoms are varied and routine care may fail to integrate menstrual history, androgen excess, laboratory findings, and ovarian imaging. Earlier recognition matters because PCOS affects fertility, metabolic health, pregnancy, psychological well-being, and quality of life. AI has therefore attracted growing interest in prediction and early detection. Relevant tools include machine-learning models for structured clinical data, deep-learning systems for ultrasound images, natural-language processing of free-text reports, and digital platforms that generate risk scores from symptoms or menstrual-tracking data. Published studies suggest promise in symptom-based triage, EHR case finding, ultrasound standardization, and emerging omics or mobile-health approaches. At the same time, literature has important weaknesses. Many studies are retrospective, single-center, or based on small, curated datasets. Diagnostic definitions are not always consistent, external validation is limited, and only a minority of studies explain how predictions would be used in real clinical practice. There are also age-specific concerns: adult criteria cannot simply be transferred to adolescents, because current guidance does not support the use of ovarian morphology or anti-mullerian hormone (AMH) as diagnostic criteria in adolescents. At present, AI should be seen as an aid to earlier case finding, triage, and standardized interpretation rather than as a replacement for clinical judgment. The most useful future systems will be the ones that are clinically understandable, fair across patient groups, validated in diverse settings, and tested in routine care pathways.
Eastern Med Coll J. January 2026 Vol.11 No.1 : 82-92
Shamsun Nahar, Musammat Shamima, Akter et al.· Eastern Medical College jour...· 0 citations
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.