If you haven’t noticed lately, we’re seeing an increase in demand for analytics driven by health reform. However, for many organizations, the culture needs to change in order to fully embrace analytics as part of the standard practice of care. Many will agree that there is too much information for clinicians to rely only on training and experience as they treat patients rather than leverage insights from analytics for clinical decision support. Providers who embrace analytics will be best positioned to improve patient care from the perspective of decreased cost, improved efficiency and enhanced patient experience.
In my role at Intel, I’m often asked where “big data” capabilities can apply to healthcare. One of the areas that always top my list are the clinical records. Roughly 70 percent of the electronic health record (EHR) includes clinically relevant information that is unstructured or in free form notes, meaning potentially critical pieces of information are not easily accessible to providers. There are many tools that can use sophisticated natural language processing techniques to pull out the clinically relevant information however, the culture has to be ready to accept those kinds of solutions and use them effectively.
Personalized medicine analytics is a combination of data bits coming from multiple data sources and each comes with its own unique set of challenges. There is the payer side, the clinical side, the biology, life sciences and genomics side and finally, the patient side and the work that we’ve been doing is in all of these areas. We look at big data and health and life sciences as the aggregation of all of these different data sources and address the challenge of how this content will be generated, moved, stored, curated and analyzed.
The goal is to take advantage of the sophisticated analytics and sophisticated technology capabilities and merge those with the changes to workflow on the healthcare side and the life sciences side and pull those two areas together to deliver care specific to an individual. This is very different from treating a large cohort of all diabetes patients or all breast cancer patients in exactly the same way.
Personalized medicine is really two different perspectives. First, is on the genomics side, where you include as an attribute to the patient care pathway the genome of that patient, comparing it against a reference genome to determine what is different about the patient as an individual or how their tumor genome differs from their normal DNA. Second, there’s the population health aspect to personalization; really understanding all of the data that is available in patient records whether it be structured or unstructured data and then developing care plans specific to that individual. For example, micro segmenting a population taking into account comorbidities and socio-economic factors with the help of advanced analytic tools.
There was a recent article in the Journal of Patient Safety that stated that there may be more than 400,000 premature deaths per year that are preventable in a hospital setting. Furthermore, 10-20 times more than that statistic cause serious harm but don’t result in death. For example, big data and analytics are being used to help identify and diagnose sepsis earlier so that it can be treated more effectively and be less costly for the payer and provider.
A great example of using wearables to better understand disease progression is the work that Intel is conducting in partnership with the Michael J. Fox Foundation for Parkinson’s research (see video above). Individuals wearing specialized devices will be tracked around the clock; observations will be recorded 300 times a second and all information will be stored in the cloud. What this means for researchers is that they will go from evaluating a few data points per month to observing 1 gigabyte of data every day.
By analyzing the existing data that is available, adding wearables, improving the velocity in analyzing data, there are a lot of opportunities to improve patient safety using some of these tools.
What questions about clinical analytics do you have? How are you using data in your practice or organization?
 James, John T. PhD. “A New, Evidence-based Estimate of Patient Harms Associated with Hospital Care.” Journal of Patient Safety (2013): http://journals.lww.com/journalpatientsafety/Fulltext/2013/09000/A_New,_Evidence_based_Estimate_of_Patient_Harms.2.aspx