Current Research and Scholarly Interests
We analyze multiple types of health data (EHR, Claims, Wearables, Weblogs, and Patient blogs), to answer clinical questions, generate insights, and build predictive models for the learning health system. Read more at http://shahlab.stanford.edu<br/><br/>We answer clinical questions to enable better medical decisions using EHR and Claims data, via a bedside consult service that enables the use of aggregate patient data at the point of care. See more at http://greenbutton.stanford.edu<br/><br/>We make predictions that allow taking mitigating actions. We characterize the fairness and examine the ethical implications of using machine learning in clinical care. We have built models for predicting future increases in cost, identifying slow healing wounds, missed diagnoses of depression and for improving palliative care. Learn more at http://aihc.stanford.edu<br/><br/>We develop methods to analyze multiple datatypes for generating insights. Such as:<br/><br/>(1) Combining molecular data with EHR data to identify biomarkers for poor outcomes in fibrotic diseases.<br/>(2) Learning effective treatment pathways in Type 2 Diabetes using medical claims data from multiple countries.<br/>(3) Monitoring Point-of-Care glucose meters using coincident testing with central laboratory measurements.<br/>(4) Detecting skin adverse reactions by analyzing content in a health social network. <br/>(5) Conducting surveillance for drug adverse events and safety of medical devices by analyzing using clinical notes in the EHR.<br/>(6) Mining Web search logs to predict health utilization and analyzing information seeking behavior of health professionals.<br/>(7) Inferring physical function from wearables data.<br/>(8) Assessing the accuracy of automatic speech recognition for psychotherapy.
