University of Louisville finds SAS(r) predictive analytics superior in ranking health care provider quality
CARY, NC (Sep. 27, 2010) - University of Louisville researchers are uncovering better ways for insurers and health care providers to comply with the US health care bill President Obama signed into law. They are using SAS Analytics software from the leader in business analytics to analyze structured, quantitative data and unstructured, textual information from medical records and health care claims. Knowledge gleaned from data and text mining can assist in claims negotiations between insurers and health care providers. Related research also defines optimal treatment plans that both improve quality of care and reduce costs - insights that health care consumers welcome.
Outdated information management strategies and invalid statistics cause serious problems in investigating health outcomes and negotiating reimbursements. Predictive modeling, however, goes beyond standard regression techniques, expanding advanced analytical options for better, faster decision making. Predictive models use a variety of tools to deliver more accurate, long-range views of treatments and costs.
Predictive analytics enable better decisions
"The truth is in the data," said Dr. Patricia Cerrito, Professor of Mathematics at the University of Louisville and author of Text Mining Techniques for Healthcare Provider Quality Determination: Methods for Rank Comparisons. "SAS' data preprocessing tools enable a thorough investigation of complex health care claims. Predictive analytics reveal relationships between treatments and outcomes, as well as costs. Many conditions that formerly required surgery, such as ulcers, are now effectively addressed through medication, which greatly reduces costs. Under the health care bill, prescription medications can appear expensive, even when they cost less than alternative treatments."
Many models attempt to determine who provides the highest quality health care with the best patient outcomes. Quality rankings have become critical for determining provider reimbursements. Cerrito has developed a patient severity index for use in predictive analytics models to rank provider quality more accurately. She will present the new model at the World Congress Leadership Summit on Predictive Analytics.
"SAS Analytics software is extremely versatile for investigating complex data," Cerrito said. Using a sample of 8 million publicly available records, she compared her severity index with public models currently in use. She discovered that health care providers using current models could boost their quality rankings without actually improving care. Yet providers delivering the best care often don't receive the level of reimbursements that reflect that quality. Cerrito believes that a comprehensive patient severity index that encompasses the entire patient record will enable accurate rankings of quality of care across providers. This will ensure that health care providers must improve quality of care to boost rankings.