Multiple-Disease Risk Predictive Modeling Based on Directed Disease Networks
This paper studies multiple-disease risk predictive models to assess a discharged patient’s future disease risks. We propose a novel framework that combines directed disease networks and recommendation system techniques to substantially enhance the performance of multiple-disease risk predictive modeling. Firstly, a directed disease network considering patients’ temporal information is developed. Then based on this directed disease network, we investigate different disease risk score computing approaches. We validate the proposed approaches using a hospital’s dataset. Promisingly, the predictive results can be well referenced by healthcare professionals who provide healthcare guidance for patients ready for discharge.
KeywordsDirected disease network Predictive modeling Multiple-disease risk assessment
A significant part of this work from Tingyan Wang and Robin Qiu was done with the support from the Big Data Lab at Penn State. This project was partially supported by IBM Faculty Awards (RDP-Qiu2016 and RDP-Qiu2017).
- 4.D.A. Davis, N.V. Chawla, N. Blumm, N. Christakis, A.L. Barabasi, Predicting individual disease risk based on medical history. in Proceedings of the 17th ACM conference on Information and knowledge management, pp. 769–778 (2008)Google Scholar
- 7.A.J. Frandsen, Machine Learning for Disease Prediction, Master thesis (Brigham Young University, 2016)Google Scholar
- 15.World Health Organization (2019) International statistical classification of diseases and related health problems, 10th Revision. Retrieved 8 Jan 2019. http://apps.who.int/classifications/icd10/browse/2016/en