Prediction of Hospitalization Cost for Childbirth
Improving cost transparency in healthcare creates opportunities to empower patients and reduce spending. Medical prices vary widely within U.S. markets but few consumers (i.e., patients) have the information to shop for their health- care services. Consumers are starting to question why their healthcare bills are so high – and why they can’t find health- care prices at all. In this work we apply predictive analytics to help consumers make educated decisions related to childbirth cost. Regression decision trees are used to rep- resent the path from the patients’ characteristics (e.g. demographics, comorbidities), choices of medical procedures, and choices of care providers, to the predicted cost variations. The outcome of this research helps consumers make informed choices as well as reduce healthcare expenditures. We demonstrate a system for childbirth cost prediction that far outperforms our baseline model in terms of prediction error. Additionally, we identify several attributes, such as labor room usage, operating room services, anesthesia, and hospital cost to charge ratio, which play a significant role in determining the overall cost of childbirth.
in: Proceedings of HI-KDD 2014 (ACM SIGKDD Workshop on Health Informatics), workshop at KDD2014 (20th ACM SIGKDD Conference on Knowledge Discovery and Data Mining)