Anxiously Awaiting the Promise of Cognitive Computing

By Scott Arnold, SVP & CIO, Tampa General Hospital

Scott Arnold, SVP & CIO, Tampa General Hospital

Healthcare IT has worked itself through a range of ideas, concepts, fads and buzzwords with mixed success in recent history. Some of the efforts are legitimate with genuine results(a tip of the hat to those claiming achievement), but more often the result is a lot of money spent with no material result (through informal polling of my colleagues). The most prevalent money pit seems to be associated with the time and expense burned on data projects that lack vision and clarity. A lot of time wasted through pulling and harvesting data, normalizing, preparing, moving, validating, fixing and handling data to cohabitate with other data in a disparate location.A mouthful and yes, expensive.

Nevertheless, data projects may have the most value to our future. I see data as a strategic asset that should be exploited to the fullest extent in healthcare (patient wellness at the center of course). Is there a solution that affords a simple data dump without all the handling and inefficiency (think single stream recycling bin)? Can I just buy the service instead of the heavy technology, consulting and labor investment? Not sure, but I am prudently hopeful for the concept and promise of cognitive computing in healthcare. The basis for my optimism is grounded in the prominent investments being made by many companies, the most notable of course being IBM (Watson Health). However, there are many working toward a similar goal. Frankly, I don’t care who is first to market as long as the solution comes sooner rather than later.

The most relevant information about a patient is usually buried in blobs of unstructured notes

With that said, here is a message to the gifted and visionary people that will turn the fantasy of cognitive computing (sans unicorns) into a reality with practical and amazing applications for the benefit of medical treatment, research, wellness, prevention, healing and acute care (to name a few):

Dear cognitive computing visionaries, inventors and developers,

I am a steward of Health IT assets and data. As a strategy, I like the idea of leveraging cognitive computing for decision support for providers and clinicians. I am encouraged by your developments toward a product that will help us make sense of the ocean of data we have accumulated for our patients. Sometimes an individual patient can have many terabytes of data in our systems to support clinicians in patient care. Even the most talented clinicians can’t possibly digest all of the data, in milliseconds with precision. Furthermore, I understand the early speculative use of cognitive computing to support oncology focusing on the computation of treatment plans measured in seconds, versus the human process measured in days or weeks. That is amazing, wow.

As you advance the concept, could you please make sure to consider unstructured data to compliment traditional structured data? To be honest, the most relevant information about a patient is usually buried in blobs of unstructured notes, but those unstructured notes ultimately provide additional context to structured data too. Also, be sure to consider a way to pull in institutional data from multiple electronic health records, biomedical device data, demographic and socio-economic data, CRM/lifestyle data, claims data, medication compliance notes, genomic data (duh, of course) and a variety of exogenous data including fitness monitors, Bluetooth devices, sensors, home monitoring equipment, patient provided notes and other data yet to be defined (basically the kitchen sink).

As a foundation, I think the idea of cataloging every medical journal and all legitimate medical reference materials as the underpinning for the master healthcare cognitive computer in the sky (as a good start). There are others smarter than me that can articulate the appropriate footing for clinical decision support, so please get a bunch of those geniuses together to help you. A word of advice, please resist the urge to amass “committees” to make decisions and slow down innovation through bureaucracy, rather use small, smart teams to help establish standards that are supported by evidence and science. Just get it out there (with reasonable and safe standards), allow the tuning to occur with “committees” later (if you must). Otherwise, I fear we may admire this problem for a long time and wait for perfection before we start using it to make better clinical decisions.

Finally, please make sure this process isn’t labor intense or expensive. I would just like to drop off my data (like a single stream recycling bin) and let the cognitive computer handle the heavy lifting. Implementation and product onboarding should painless, unlike an EHR implementation please! Also, consider some of us that are in the small to mid-sized institutions as partners in your development, instead of the usual list of large, big name institutions. You would be surprised how much more quickly some of us can move (as opposed to the huge, big name places) with as much or more motivation. In terms of a business model, please make it easy for me to buy this from you as a service, and please find a way to make it affordable (even for individual providers). I know that is a lot to ask, but I am confident in the talent rallied around this cause for the benefit of the human race. Anxiously awaiting your developments. Please hurry.