Improving Healthcare Outcomes with Business Analytics
Majidi Zolbanin, Hamed
MetadataShow full item record
This work includes three studies, in which we develop the fact-based data analytics methodologies to inspect, clean, transform, and model data to discover useful information and make suggestions that enable enhanced decision making in the healthcare sector. We employ this methodology to analyze medium and large physical and mental health data to address questions that could not be easily answered using the traditional methodologies. In the first study, we show that the availability of comorbidity data enables more accurate predictive models, which results in better decision making in future. This can be done with minimum changes in existing data storage systems, as most methods of recording comorbidities provide a single score to account for overall health of patients.In the second study, we build a predictive analytics model that can inform drug treatment court authorities to make more efficient decisions by favoring offenders that have greater chances of completing the treatment. Our results confirm prior research and add to it by creating a predictive model that could separate completers from dropouts with an accuracy over 90%. We also show how predictive analytics provides an explanation for the behavior of offenders in these programs from a different angle. At the end, we provide a list of factors and the values they take to enable an evidence-based decision making in these treatment programs.The third study uses a subset of the Cerner Corporation�s massive data warehouse to explore how hospitals have responded to the Hospital Readmission Reduction Program (HRRP) of the Patient Protection and Affordable Care Act. We use the ecological rationality framework of rational choice theory to show that hospitals selected an outcome-oriented approach to lower readmission rates, that is, through fewer inpatient admissions. We also employ a predictive approach to build models that can specify 30-day readmission propensity of patients.
- OSU Dissertations