Program: Data Science Master's
Location: Not Specified (remote)
Student: Pouria Niknam
This study investigates the risk factors contributing to early readmission among patients with diabetes. Diabetes is a common chronic disease in the US and one of the significant causes of death. Direct and indirect costs related to diabetes exceed $300 billion. Diabetes also leads to many other complications, such as cardiovascular conditions. Studies have shown that readmission rates among patients with diabetes are significantly higher, resulting in a lower quality of life for the patient and additional costs and burdens to the healthcare system. In addition, studies show that different patient support programs tailored for patients with diabetes help significantly lower the readmission rate. My objective was to find critical factors contributing to early readmission that hospitals can use to create patient support. I created two models with similar accuracy and sensitivity using logistic regression, and Elastic Net penalized repression. However, the penalized regression model has fewer coefficients, making it a more efficient choice. The results indicate that being older than 60, having a cardiovascular condition, and having repeated inpatient and emergency visits in a year are the most critical factors resulting in early readmission. Therefore, patient support programs that address overall health for patients older than 60, especially those with diabetes and cardiovascular conditions, may significantly reduce early readmission.