Predicting Patient Outcomes During and After Hospitalization Using - PowerPoint PPT Presentation

Predicting Patient Outcomes During and After Hospitalization Using AI Aziz Nazha, MD Director, Center for Clinical Artificial Intelligence Associate Medical Director, Enterprise Analytics Assistant Professor of Medicine Lerner College of Medicine / CWRU Taussig Cancer Institute Cleveland Clinic @AzizNazhaMD

Google Brain 216,221 46,864,534,945

Google Brain

Model Building ü 1 ü 2 EMR Data using Algorithm Variable Importance Final Model NLP

Outcomes During Hospitalization (CCAI) 1,485,880 Hospitalization 708,089 Unique patients Between 1/2011- 5/2018 Abbreviations: ROC = Receiver Operating Characteristic, AUC= Area Under the Curve, RMSE = Root mean squared error

Our Model Explainability for 30-days Hospital Readmissions

Our Model Explainability for Gender

Personalized Explanation of the Model Output

Conclusions ü We build a personalized prediction model for hospital outcomes ü We used AI algorithm to learn something new

Cleveland Clinic Every Life Deserves World Class Care. E-mail: nazhaa@ccf.org