Beware of Hits, Misses of Data to Transform Care
In the journey to value-based care, developing a robust healthcare analytics foundation to support transformational change is imperative, with successes but also many challenges along the way.
In the journey to value-based care, developing a robust healthcare analytics foundation to support transformational change is an imperative, with successes but also many challenges along the way.
That was the message from Graham Hughes, M.D., CMO at SAS Center for Health Analytics and Insight, during the July 14 opening keynote at the Healthcare Analytics Symposium in Chicago, sponsored by Health Data Management. Hughes started by giving an example of how a good idea can quickly turn sour, based on a February 2012 article in the New York Times.
Retailing giant Target started analyzing customer buying patterns and among its findings learned that it could predict with high certainty women who were pregnant, particularly in the third trimester, based on what they were buying. The retailer then would send congratulations to these consumers as well as store vouchers. They sent communications to a teen who had not yet told her parents about her pregnancy, and that program was quickly shut down.
Another challenge is the use of electronic health records data for population health management and patient engagement, among other uses. But EHR coding carries a 20 percent error rate, Graham said, and 80 percent of EHR data is in free text. Consequently, organizations have to question the veracity of the data they have and determine how much uncertainty can be tolerated.
Many healthcare organizations do not have the resources to conduct robust analytics by themselves and should consider collaborating with other stakeholders, Graham advised. Were starting to see data collaborations breaking out all over the place. Further, the healthcare industry quickly will learn what other types of businesses already have--that they need more power and speed to perform analyses in two minutes, not four hours.
Graham also gave examples of healthcare organizations improving processes through data mining. Duke University Health System analyzed NICU workflows to made changes that reduced mortality rates. Mayo Clinic is text mining discharge records to automatically detect if patients sent home are following up on appointments.
That was the message from Graham Hughes, M.D., CMO at SAS Center for Health Analytics and Insight, during the July 14 opening keynote at the Healthcare Analytics Symposium in Chicago, sponsored by Health Data Management. Hughes started by giving an example of how a good idea can quickly turn sour, based on a February 2012 article in the New York Times.
Retailing giant Target started analyzing customer buying patterns and among its findings learned that it could predict with high certainty women who were pregnant, particularly in the third trimester, based on what they were buying. The retailer then would send congratulations to these consumers as well as store vouchers. They sent communications to a teen who had not yet told her parents about her pregnancy, and that program was quickly shut down.
Another challenge is the use of electronic health records data for population health management and patient engagement, among other uses. But EHR coding carries a 20 percent error rate, Graham said, and 80 percent of EHR data is in free text. Consequently, organizations have to question the veracity of the data they have and determine how much uncertainty can be tolerated.
Many healthcare organizations do not have the resources to conduct robust analytics by themselves and should consider collaborating with other stakeholders, Graham advised. Were starting to see data collaborations breaking out all over the place. Further, the healthcare industry quickly will learn what other types of businesses already have--that they need more power and speed to perform analyses in two minutes, not four hours.
Graham also gave examples of healthcare organizations improving processes through data mining. Duke University Health System analyzed NICU workflows to made changes that reduced mortality rates. Mayo Clinic is text mining discharge records to automatically detect if patients sent home are following up on appointments.