Healthcare is a huge and growing segment of our economy, and represents enormous potential as an area for both cost reduction and improvement in quality of care. A particular area of interest is hospital readmission rates (i.e., needing to return to the hospital soon after an initial treatment there), which the federal government plans to begin tracking soon, and which many hospitals are doing on their own already. That, along with all of the potential information (much of it unstructured) that could relate to the reasons for those readmissions, represents both a big challenge but an even greater opportunity for the BI community. Being able to discover those factors that correlate with readmissions would enable providers to identify high risk patients, and take the steps needed to reduce readmission rates.
A recent article by Tom Groenfeldt in Forbes magazine, “Big Data Delivers Deep Views Of Patients For Better Care “, discusses a company, Seton Health Care Family, already involved in an analytics project focusing on the use of big data to improve patient outcomes. They have been dealing with the complexity of “more than two million patient contacts a year, usually attended by dozens of existing patient record pages and generating more with each visit — some electronic, many notes in physician’s infamous handwriting, some transcribed from dictation and others jotted down on X-rays”.
The Future of Big Data
With some guidance, you can craft a data platform that is right for your organization’s needs and gets the most return from your data capital.
Working with an analytics tool from IBM, Seton started with a pilot project on patients with congestive heart failure. This is not the easiest condition to measure, because a poor outcome could be immediate, or it could take ten years to show up. The challenge was how to deal with the 80% of patient information that is unstructured, which includes not just physician’s notes, e.g., but also patient information that would not be in their medical records, such as a social infrastructure or family support system to help make sure they know what to do with their care plan and their medications, or whether they had access to transportation for checkups or treatments.
Even if some of this information might be stored somewhere in free text, it is often be hidden in mounds of paper. In other cases, it might require a more detailed patient interview or even actually going to their home to gain a full understand of these “off the medical chart” factors. So, to truly determine the risk factors most highly correlated with a return hospital visit, more of this unstructured data needed to be gathered, it needed to be transcribed, and it needed to be stored and organized. Only then, can a full analytics process begin. The preliminary results at Seton have surprised the staff, because the results suggest that it is those outside (unstructured) factors, rather than lab values or clinical measures, which seem to have the biggest impact on readmission rates.
If this conclusion holds, and bigger predictors of successful outcomes are based more on these personal circumstances/external factors, than on what actually goes on inside the hospital on that initial stay, that could require changes in health care beyond simply tightening up or improving internal hospital procedures.
But, we are still in the beginning stages of this cycle, and first step in that process is gathering more data, and getting into an easily readable and digestible format that can be properly evaluated through the use of a BI analytics tool. In Part II of this blog, we will discuss further conclusions that can be drawn from this, as well as exploring another real use case in the health care field.
