101,766 clinical records from 130 US hospitals documenting diabetic patient encounters between 1999 and 2008. The data contains 50 features covering patient demographics, hospital stay duration, and 24 specific medication adjustments.
Use Cases
- Predict the 'readmitted' status using 'num_medications' and 'time_in_hospital' as input features
- Analyze the correlation between 'A1Cresult' and the frequency of 'number_emergency' visits
- Evaluate the impact of medication 'change' on the likelihood of early hospital readmission
Strengths
- 101,766 rows of clinical data spanning a 10-year period
- Includes 50 features such as 'time_in_hospital', 'num_lab_procedures', and 'number_diagnoses'
- Tracks 24 different diabetes medications and dosage changes via the 'change' column
- Target variable 'readmitted' is categorized into three classes: <30 days, >30 days, and No