What statistical methods and models are supplied?
Response: We have internal access to millions of de-identified hospital records in both the inpatient and outpatient settings. These training/test data are key to addressing the predictive analytics demands of clients and site customization. Data modeling and algorithm development are performed using industry-leading tools for data mining and supervised machine learning such as Weka, Orange, Spotfire, and R. Ongoing efforts include decision tree and regression models for a generalized predictor of hospital readmissions using variables such as length of stay, age at admit, APRDRG severity of illness, discharge day, discharge time, total number of admissions, etc. Our regression based global readmissions algorithm currently approaches 80% accuracy. This is a significant metric to forecast, and accuracy will improve as specific patient populations, such as heart failure, continue to be targeted.