event-icon
Description

Hospitalizations are resource consuming for hospitals, insurers, and society. Leveraging large amount of medical claims data, we developed a predictive model for future hospitalization using neural network. Without using any expert domain knowledge, our model performance is competitive to models that utilized expert domain knowledge, such as Clinical Classifications Software (CCS). Modern machine learning techniques might be leveraged to reduce manual efforts in data mining and speed up insights discovery.

Learning Objective 1: We explored applying neural network, a modern technique that can leverage a large scale of complex data, on pediatric medical claims and developed a competitive predictive model to predict patient’s next year’s hospitalization without any expert domain knowledge.

Authors:

xianlong zeng (Presenter)
Ohio University

Soheil Moosavinasab, The Research Institute at Nationwide Children’s Hospital
Enju Lin, The Research Institute at Nationwide Children’s Hospital
Yungui Huang, The Research Institute at Nationwide Children’s Hospital
chang liu, Ohio University
Simon Lin, The Research Institute at Nationwide Children’s Hospital

Presentation Materials:

Tags