Every night on the evening news your local weatherman makes a forecast that combines technology, weather models and climate measurements. In the corporate world, publicly-traded corporations spend considerable time and effort making financial projections comparing past business performance with current market conditions. On the political front, campaigns spend millions of dollars on polling to gauge how particular populations plan to vote in the future.
In short, predictive analytics are all around us. And as the digital age progresses, using data to forecast outcomes will be used even more across the industrial landscape, including healthcare.
But, how do you look at healthcare data to predict future clinical outcomes?
Right now we can’t for the simple reason that we don’t have enough healthcare data to analyze yet. It is entirely possible, however, that within the next few years pioneering technology and population health management companies will refine their processes and be able to forecast clinical outcomes with certainty.
Within the massive healthcare complex right now, we’re still largely in a position of aggregating data and getting siloed information into one data lake. Data intelligence analytics are slowly being applied, which in essence fill in data gaps and enable data modeling. Algorithms are being incorporated, and through trial and error, improved.
As we get more of this data cleaned up and analyzed, we will be able to make more accurate projections. Only a few years into this process, clinical predictions can be made at around 60% certainty.
To go out into the market and tell providers that projections are signaling that they will have to contend with a particular situation in the not-too-distant future, you need to be around 80%-90% certain.
And once data scientists close in on that 80%-90% rate, using data to make healthcare projections will be a limitless growth opportunity.
For now, however, current research appears to focus on diagnosis and procedure codes and demographics. Soon enough, lab results, medications, vital signs, and softer elements like patient mood and pain levels will also be incorporated to strengthen projections. In time, if there is enough data, we’ll be able to say these protocols for this group of people with these types of behaviors are working well…and for another group of people with a higher pain level, other protocols will work.
Further, as the healthcare industry continues to employ more holistic approaches toward patient care, social determinants of health - factors like socioeconomic status, language barriers and access to transportation – can be used to enhance the predictive modeling process.
So as predictive analytics expand in healthcare, there will be enormous opportunities for growth. While we’re still a few years away from full implementation, tremendous progress is being made.