Researchers from MIT have revealed that health knowledge graphs which are used to help clinicians with the medical diagnosis are ineffective for certain conditions and patient groups. Researchers analyzed various models that use electronic health record (EHR) data, containing medical histories of patients, to identify the relation between the disease and symptoms. They observed a huge discrepancy in the data of models for diseases that have high percentages of very old or young patients. However, they believe choosing the appropriate data and model with some amendments can improve performance. The research would be helpful to develop accurate tools to assist physicians and patients in medical diagnosis and to find accurate relationships between diseases and symptoms.
