A Method and Practice for Menopausal Disease Prediction Based on Knowledge Graph

As the population of women entering menopause grows, so does the potential for chronic conditions like os-teoporosis, cardiovascular disease, and metabolic syndrome, which underscores the the critical need of early intervention and treatment during menopause. Meanwhile the Electronic Healthcare Records (EHR) is becoming a valuable resource for health event predictions, including diagnosis, mortality, length-of-stay, and readmission. However, effectively utilizing diagnosis features and co-occurrence relationships among medical concepts in EHR data remains a challenge. To address this issue, we propose a new knowledge graph-based approach, which can integrate co-occurrence relationships among medical concepts and numerical information in EHR data. Our approach leverages a concurrency-aware graph learning method and a feature-wise fusion block to enhance knowledge representations and fully exploit the information in patients' EHR data. Experiments using actual EHR data showcases a superior efficacy of our model in forecasting menopausal diseases compared to existing approaches.