Improving Healthcare with Data Analysis and Predictive Modeling
The rapid changes in not only the financing of health care, but also the health and wellbeing of the patients we care for are driving the need to better understand and care for patients. One of the keys to providing more effective, valuebased care is to understand the population of people we serve.
Our challenge is to move beyond the collecting of the data around our patients, which is what the industry has been doing for the last 20 years, to truly understand that data in order to drive strategies around more effective care to keep patients healthier. IT is the interface of the data, delivering to patients and employees in a digestible and user-friendly way.
A few years, Carolinas HealthCare System recognized the need to harness the data it collects on our patients to better understand and develop new methods for treatment and care. Our analytics center was formed to begin analyzing data on our patients that provides clues to improving care, understanding our population’s health and provides for predictive analytic work.
For instance, according to the Centers for Disease Control and Prevention, nearly 65 percent of all adults in North Carolina are overweight with 28 percent being obese; and nearly a third of our children are overweight or obese. This rapid rise in obesity in our population has changed the type of care being sought by our patients and it’s changed the conditions and complexity of conditions being treated by our clinicians.
Here at Carolinas HealthCare, we understand through our data analysis and predictive modeling that obesity, diabetes, asthma and other conditions require and we improve the care being delivered.
In order for us to understand how to better deliver this care, we did more than just create a Big Data repository (although we did do that). We also committed the resources to analyze and synthesize that data into relevant stories that can help us improve the care delivered.
“Big Data” is the buzzword de jour in health care. Merely mining data does not benefit the outcome for a patient if a hospital, or system, lacks effective capability to analyze the information and understand best next steps.
We took the Big Data to the next level. For instance, a number of our hospitals are analyzing up to 40 (patient) data points related to readmission risk every hour, leading to better point of care decisions by our physicians. We’re providing better outcomes, better advising our patients of matters related to their health, and reducing readmissions because our data are mined with purpose, and analyses drive more effective delivery of care.
This model for predictive readmissions is an innovation already impacting the level of care we are providing. Before a patient even leaves one of our hospitals, our model analyzes both Electronic Medical Record data and administrative data to determine the patient’s risk of being readmitted within 30 days.
Our providers can predict, with nearly 80 percent accuracy, a patient’s risk for 30- day readmissions. Our model also identifies the key individual factors about each patient that might increase their readmissions risk allowing them to use personalized evidence-based interventions to drastically reduce the chance of readmission.
"The value isn’t the volume of data you store; the value is in understanding how the data can provide clues to improved care"
But we go beyond looking at the data we hold on our patients. Through the adoption of capabilities that allow us to also access patient data from health systems and hospitals across the nation, we are able to achieve an even better view of clinical and financial opportunities and benchmark our performance against those other systems.
Analyzing and synthesizing the data to give a better and truer picture of our patients’ health status provides us with the pathways to deliver better care with improved efficiencies. At the same time, the data helps clinicians set goals for their patients and assist them in building the support structures needed for the patient to make appropriate behavior changes.
The value isn’t the volume of data you store; the value is in understanding how the data can provide clues to improved care.