Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach

Abstract Poly-cystic ovary syndrome (PCOS) is a prevalent condition that impinges women in their prime reproductive years. Its primary trait is the elevated levels of androgen, a male hormone in female body and leads to several symptoms, including ovarian cysts, irregular menstruation cycles, infert...

Full description

Saved in:
Bibliographic Details
Main Authors: S. Reka, T. Suriya Praba, Krishna Kumar Manchala, Anna Venkateswarlu
Format: Article
Language:English
Published: Springer 2025-07-01
Series:International Journal of Computational Intelligence Systems
Subjects:
Online Access:https://doi.org/10.1007/s44196-025-00931-3
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849331827902251008
author S. Reka
T. Suriya Praba
Krishna Kumar Manchala
Anna Venkateswarlu
author_facet S. Reka
T. Suriya Praba
Krishna Kumar Manchala
Anna Venkateswarlu
author_sort S. Reka
collection DOAJ
description Abstract Poly-cystic ovary syndrome (PCOS) is a prevalent condition that impinges women in their prime reproductive years. Its primary trait is the elevated levels of androgen, a male hormone in female body and leads to several symptoms, including ovarian cysts, irregular menstruation cycles, infertility, obesity, and excessive hair development. Since PCOS exact cause is unknown and its symptoms are uncertain, a timely and accurate diagnosis is essential. In these situations, a (ML) Machine Learning-based PCOS prediction model aids in the diagnostic procedure, addresses time constraints and potential inaccuracies. A plethora of feature selection techniques have been built to discover the best optimal features for the classification of medical dataset. However, finding an efficient solution is still difficult due to noise and redundant information which may degrade the model performance. In this article, a hybrid filter-wrapper approach is proposed to identify the optimal attributes. An ensemble filter method is built, the union of top k attributes from individual filters is considered for further process. Then, Reinforcement Learning-based Binary Equilibrium Optimizer is used to find the reduced optimal features. Here, RL uses SARSA (State–Action–Reward–State–Action) to increase the population diversity in the search space, to avoid the problem of local optima, finally balances the exploration capability during the search. Then, identified optimal features are given as input to BMFK (Bonferroni Mean Fuzzy KNN) classifier and achieved 96.32% of accuracy. Further, several machine learning classifiers have been employed for the prompt diagnosis of PCOS. Finally, Random Forest has produced the highest accuracy of 95.62% than other models.
format Article
id doaj-art-6b089991200042dfa6861b47d1680c19
institution Kabale University
issn 1875-6883
language English
publishDate 2025-07-01
publisher Springer
record_format Article
series International Journal of Computational Intelligence Systems
spelling doaj-art-6b089991200042dfa6861b47d1680c192025-08-20T03:46:24ZengSpringerInternational Journal of Computational Intelligence Systems1875-68832025-07-0118112310.1007/s44196-025-00931-3Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization ApproachS. Reka0T. Suriya Praba1Krishna Kumar Manchala2Anna Venkateswarlu3School of Computing, SASTRA Deemed UniversitySchool of Computing, SASTRA Deemed UniversitySchool of Computing, SASTRA Deemed UniversitySchool of Computing, SASTRA Deemed UniversityAbstract Poly-cystic ovary syndrome (PCOS) is a prevalent condition that impinges women in their prime reproductive years. Its primary trait is the elevated levels of androgen, a male hormone in female body and leads to several symptoms, including ovarian cysts, irregular menstruation cycles, infertility, obesity, and excessive hair development. Since PCOS exact cause is unknown and its symptoms are uncertain, a timely and accurate diagnosis is essential. In these situations, a (ML) Machine Learning-based PCOS prediction model aids in the diagnostic procedure, addresses time constraints and potential inaccuracies. A plethora of feature selection techniques have been built to discover the best optimal features for the classification of medical dataset. However, finding an efficient solution is still difficult due to noise and redundant information which may degrade the model performance. In this article, a hybrid filter-wrapper approach is proposed to identify the optimal attributes. An ensemble filter method is built, the union of top k attributes from individual filters is considered for further process. Then, Reinforcement Learning-based Binary Equilibrium Optimizer is used to find the reduced optimal features. Here, RL uses SARSA (State–Action–Reward–State–Action) to increase the population diversity in the search space, to avoid the problem of local optima, finally balances the exploration capability during the search. Then, identified optimal features are given as input to BMFK (Bonferroni Mean Fuzzy KNN) classifier and achieved 96.32% of accuracy. Further, several machine learning classifiers have been employed for the prompt diagnosis of PCOS. Finally, Random Forest has produced the highest accuracy of 95.62% than other models.https://doi.org/10.1007/s44196-025-00931-3Poly-cystic ovary syndromeMachine learningReinforcement learningFeature selectionSARSA
spellingShingle S. Reka
T. Suriya Praba
Krishna Kumar Manchala
Anna Venkateswarlu
Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
International Journal of Computational Intelligence Systems
Poly-cystic ovary syndrome
Machine learning
Reinforcement learning
Feature selection
SARSA
title Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
title_full Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
title_fullStr Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
title_full_unstemmed Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
title_short Comprehensive Feature-Driven PCOS Predictor: A Reinforcement Learning-Based Binary Equilibrium Optimization Approach
title_sort comprehensive feature driven pcos predictor a reinforcement learning based binary equilibrium optimization approach
topic Poly-cystic ovary syndrome
Machine learning
Reinforcement learning
Feature selection
SARSA
url https://doi.org/10.1007/s44196-025-00931-3
work_keys_str_mv AT sreka comprehensivefeaturedrivenpcospredictorareinforcementlearningbasedbinaryequilibriumoptimizationapproach
AT tsuriyapraba comprehensivefeaturedrivenpcospredictorareinforcementlearningbasedbinaryequilibriumoptimizationapproach
AT krishnakumarmanchala comprehensivefeaturedrivenpcospredictorareinforcementlearningbasedbinaryequilibriumoptimizationapproach
AT annavenkateswarlu comprehensivefeaturedrivenpcospredictorareinforcementlearningbasedbinaryequilibriumoptimizationapproach