Poster: Mobile Menstrual Health Advising with Multimodal Feature Engineering

Menstrual pain has constituted a persistent public health challenge, impairing women's socioeconomic participation through recurrent functional limitations. Conventional diagnostic methodologies relying on symptomatic retrospection demonstrate inadequate prognostic capability for anticipatory care planning. To address these challenges, we present PainCare, an intelligent predictive system integrating multimodal self-reported data with efficient language model processing for menstrual pain forecasting. PainCare employs an advanced feature engineering process to maintain performance across varying pain severities, demonstrating significant improvements in pain prediction. Our validation study with 176 participants achieved 94.7% prediction accuracy across four pain intensity levels, a 43.5% improvement over conventional methods.