Xiaoran Tong, Ph.D., Assistant Professor, Center for Innovation in Population Health, University of Kentucky

"A machine learning assisted process to build interpretable decision support models from large data: standardize the decision-making in child welfare. "

Child welfare workers have been using standardized metric "Child & Adolescent Needs and Strength" (CANS) to capture the state of wellbeing of the youth under care. Experts have been building decision support models (DSM) algorithmically advising the best course of action on the children according to their CANS. We aim to automate the crafting of decision support model via machine learning (ML) of large data accumulated in the welfare system. The proposed automation took advantage of ML, so the DSM can accurately predict the "would be" decision of human experts with years of clinical knowledge in foster care, while addressing the often-criticized lack of interpretability due the "black box" surrounding a ML derived predictive model.

Stay connected TwitterFacebook LinkedIn YouTubeInstagram