Scholarlink

Despite growing recognition of the menstrual cycle’s influence on female athletic performance, personalized, data-driven solutions remain scarce in mainstream sports technology. This paper introduces CycleAware, a novel machine learning-powered web platform that bridges the gap between menstrual health and athletic optimization. By augmenting a real-world dataset to ensure balanced representation across all menstrual phases, we enable accurate modeling of nuanced physiological trends. A deep-tabular hybrid architecture, TabNet for interpretable feature learning combined with Random Forest for robust classification, was developed and benchmarked against a classical SVC with PCA pipeline. The hybrid model’s ability to autonomously prioritize key features while capturing complex interdependencies results in significantly improved predictive accuracy and generalization. Integrated into a user-centric Django web application, CycleAware empowers athletes and coaches to log cycle and training data, visualize historical trends, and receive real-time, phase-aware performance insights. This work pioneers a scalable, AI-integrated approach to menstrual cycle prediction, promoting inclusivity and data-informed training in competitive sports - a space where personalization and female physiology have long been overlooked.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CycleAware: A Machine Learning Framework for Menstrual Phase Prediction and Athlete Performance Tracking

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shreya Srikant,
  • Shreya Naveen,
  • Aditya Verulkar,
  • V. Shrujan,
  • N. Chaitra

摘要

Despite growing recognition of the menstrual cycle’s influence on female athletic performance, personalized, data-driven solutions remain scarce in mainstream sports technology. This paper introduces CycleAware, a novel machine learning-powered web platform that bridges the gap between menstrual health and athletic optimization. By augmenting a real-world dataset to ensure balanced representation across all menstrual phases, we enable accurate modeling of nuanced physiological trends. A deep-tabular hybrid architecture, TabNet for interpretable feature learning combined with Random Forest for robust classification, was developed and benchmarked against a classical SVC with PCA pipeline. The hybrid model’s ability to autonomously prioritize key features while capturing complex interdependencies results in significantly improved predictive accuracy and generalization. Integrated into a user-centric Django web application, CycleAware empowers athletes and coaches to log cycle and training data, visualize historical trends, and receive real-time, phase-aware performance insights. This work pioneers a scalable, AI-integrated approach to menstrual cycle prediction, promoting inclusivity and data-informed training in competitive sports - a space where personalization and female physiology have long been overlooked.