• Analyzing propensity of hospital readmissions of diabetic customers to reduce medical expenditure

      Bandi, Ravi; Pokhriyal, Shitanshu; Khan, Shakeel A. (2021)
      Background: A large portion of hospital inpatient management expenditure is due to high readmission rates. Diabetes is one of the leading causes of re admissions for chronically ill patients. Analyzing readmission patterns helps proactively manage and reduce readmission, thus resulting in reduced medical expenditure. Objective of the study is to find factors that lead to readmission of Diabetic patients and identify key influencers impacting readmission rates. Study Design Methods: The data originated from Cerner EMR systems with instances for over 70,000 patients and has information on Inpatient admission, diabetes type, length of stay, Lab tests performed, and medications administered across 130 US hospitals We used Logistic Regression, Na�ve Bayes and Classification tree methods and Data visualization by using key influencers to identify the key factors. Results & Findings: Using the above data mining and visualization techniques, the study had key findings as below: - Outpatient Diabetics above age 40 with HBA1c level more than 8 are more likely to readmit. - Inpatient diabetic patients with higher number of prescribed medications and number of procedures are less likely to get readmitted. - Serum level analysis shows that with High glucose serum levels, the readmission rates are higher. - Higher HbA1c in patients has a direct relationship with re admissions. - Lab Procedures show high correlation with medications, diagnosis, and time in Hospital.