Medicare spends $26 billion annually on hospital readmissions and because of
this there are many campaigns to reduce the rate of hospital readmissions
(Wilson, 2019). The data analytics problem that I am analyzing is which
factors impact hospital readmissions. I plan to develop a model for predicting
early readmission (within 30 days) which can help hospitals identify patients at
risk of readmission and potential ways of modifying treatment plans, outreach,
and follow-up. Reducing the readmission rate reduce the resource burden on
the healthcare system (i.e., nurses, doctors, beds, etc.) and the financial burden
on the patients.
Several studies have already been done to develop models for assessing a
patient’s risk of readmission or not showing up to a scheduled appointment.
Predicting hospital readmission has been explored by healthcare data scientists
on multiple subsets of patients. For example, Andrew Long (2019) and Min et
al. (2019) have attempted to develop predictive models on subsets of diabetic
and COPD patients, respectively. However, both Long and Min et al. produced
models with AUC of approximately 0.65. I intend to improve upon what has
already been done by developing a model with better performance.
References
Long, A. (2020, January 30). Using machine learning to predict hospital
readmission for patients with diabetes with Scikit-Learn. Medium. Retrieved
May 31, 2022, from https://towardsdatascience.com/predicting-hospital-
readmission-for-patients-with-diabetes-using-scikit-learn-a2e359b15f0
Min, X., Yu, B., & Wang, F. (2019, February 20). Predictive modeling of the
hospital readmission risk from patients' claims data using Machine Learning:
A case study on COPD. Nature News. Retrieved May 31, 2022, from
https://www.nature.com/articles/s41598-019-39071-y
Wilson, L. (2019, June 26). Ma patients' readmission rates higher than
traditional Medicare, study finds. Healthcare Dive. Retrieved May 31, 2022,
from https://www.healthcaredive.com/news/ma-patients-readmission-rates-
higher-than-traditional-medicare-study-find/557694/
