I needed a few things from Home Depot to repair my lawn’s sprinkler system last weekend. So, I made a reservation to rent an 18-wheeler for six months from now, applied for a government grant to fund it and paid for some consultants (up front) to show me how to use it. The truck is bright red and has a cool airhorn. It looks like a Transformer and should be awesome. My grass will probably be dead by then though and I probably can’t afford the water bill next summer now that I think about it.
This crazy story is a metaphor for how most health systems are approaching Big Data.
It’s all too familiar that Big Data comes at “Big Cost”. Big Data has the potential (and in some cases the track record) to be a catalyst behind increasingly efficient and higher-quality healthcare – let’s call it “enlightened healthcare”. The demand for enlightened healthcare exists across thousands of providers regardless of setting or size. Making these Big Data tools affordable is essential to the spread of enlightened health care and health systems are spending a fortune on Big Data in pursuit. In the context of a healthcare macro environment that still needs to shed at least 20% of the cost, this makes no sense.
Despite health systems funding a thriving Big Data industry segment through hundreds of millions of dollars invested (just go see the booths at HIMMS!), the average hospital still does not get undeniable ROI from Big Data. And the average hospital is certainly not ready to add artificial intelligence into the milieu. The competent generalists staffed by the average health system have not proven they can quickly deploy the right amount of technology for the problem nor consistently organize and drive internal performance improvement projects. These efforts typically result in committees, multiple vendor evaluations, 18-month projects and laborious debate.
Big Data empowering enlightened healthcare soon becomes intertwined with population health and other macro-initiatives. And with that comes the quite mutual assent that this will take years. It can’t be helped we say as we push back from the meeting table.
Meanwhile, patients, real people with problems, are being run through care processes that miss diagnoses, skip follow-up and have too much variation. Some of them will die because of it. And care providers, real people solving problems, trudge along making the best decisions they can without the full decision support their organization is most likely already capable of delivering. Some of them will quit because of it.
We need technology that disrupts the “$3 M and 3-year data warehouse” paradigm. We need the meta-data for each patient or process sufficient to be surveilled, but not necessarily a data warehouse. We need an advanced but adoptable means of real-time alerting and retrospective analysis by individual patient care process. We need a tool that can take advantage of the data already in the hospitals’ systems to analyze structured AND unstructured data for the missing care or risk. We need it to be implemented in ninety days and cost less than six figures.
Instead of the prevalent “Big Data” model, why not approach this from a lean perspective? Why not address, monitor and alert the five patient care processes the hospital already knows need improvement? Why not produce the just-in-time metadata sufficient to go after these problems? Why not buy just enough technology to get an ROI, then build on the success?
For example, incidental findings from outpatient medical imaging are a persistent problem. The findings from radiologists (doing their jobs and documenting further follow up needed) often sit unread and wait for manual handling simply because the language around the finding is unstructured and buried in a lot of other information. Fixing this “miss” has been shown to lead to early-stage cancer detection repeatedly. How much more of a call to action is required? And yet the approach to fixing this with readily and economically available technology is often buried in larger efforts around Big Data. So, patients wait and health systems underperform while we wait for our Big Data warehouse.
If you have read this far, it is certainly self-serving for me to now state that we have such a tool at iWT health called Notifi. But it makes my point no less valid. Notifi was developed and applied to solve the example above regarding incidental findings for several years now. As the failed Big Data approach became clear to us iWT health, Notifi was enhanced and purposed to solve other patient care misses and surveillance too.
Regardless of whether you find NOTIFI or some other tool that works, health systems should reconsider the conventional wisdom. The “$3 M and 3-year data warehouse” option is not working for most and in fact, Big Data becomes an excuse for delays, failures and wasted resources. Meanwhile, patients and caregivers pay a horrible price.
Go “little” on Big Data. Jump in your car and go get what you need. Forget the 18-wheeler. Air horns are overrated.