With the Christmas ham consumed, the eggnog drained, and nothing left of the cookies but the crumbs, perhaps some of us here are feeling a tad ill today. Let’s hope those digestive pangs are nothing more than a reminder of a holiday well spent.
As cheery as a hospital might try making itself during the holiday season, it’s really no place to visit, let alone stay. It’s too bad just thinking about healthcare has the potential of turning so many stomachs these days.
Healthcare promises to be one of the most discussed and watched industries in 2014, as patients, providers, and insurers continue negotiating their way through all that the Affordable Care Act has wrought. I could get mired in the negatives, but I’ll stay positive here.
Healthcare in the US needs change, and that’s coming on the wings of better data, advanced analytics, and innovative thinking.
Let’s head to Indiana, where some interesting healthcare initiatives are afoot. Earlier this year, I talked with Curt Sellke, who was then vice president of analytics at the Indiana Health Information Exchange (IHIE). In From Bad Accident to Good Analytics, Sellke had shared his plans to use predictive modeling to help prevent hospital readmissions, and I thought it’d be interesting to touch base with him for a status report.
Sellke is now vice president of analytics at Thrive HDS, a for-profit IHIE subsidiary that launched Oct. 1. “We’re taking all that we’ve learned and the technologies we’ve developed in Indiana and going beyond the state borders,” he said. “From an analytics perspective, the Indiana client base has provided a great learning lab for us.”
As I’d discussed in that early post, the IHIE offers a service that alerts accountable care organizations (ACOs) and hospitals in the state when a patient is admitted, discharged, or transferred. It’s able to do so in part because it maintains what is considered one of the country’s largest inter-organizational clinical data repositories, the Indiana Network for Patient Care (INPC). This database holds more than 20 terabytes of data and 10 million unique patient records.
Now, it’s just starting to test the addition of a predictive score to that alert, taking it “to the next level,” Sellke said. Thrive HDS is feeding patient data from the INPC into a predictive analytics model created using the Predixion Software analytics platform. “We’re going to tell them what we think the possibility of readmittance is and what are the three or four key drivers for that score, plus what is it about that patient that may be driving them to a higher risk than others and where they fit into the stratification — high, medium, low.”
Full Article
