Thursday, January 08, 2015

In pursuit of the big data Holy Grail

About every six months, one of the young brilliant and enthusiastic "big data" people who hang out in the MIT neighborhood near Kendall Square or in the Bay Area of California comes to me for advice as to how to break into the health care market.  He or she is inevitably prepared to deliver the Holy Grail to a waiting health care world, i.e., a real-time decision support system that will codify the world of evidentiary medicine and help clinicians reduce length of stay, the number of unnecessary readmissions, and the cost of care.  The person has sometimes, but not always, set up a "comparable" company in another field, analyzing big data and improving industrial processes, and s/he has often sold that business for a handsome sum to a multinational corporation or private equity firm.

I love meeting with these young people.  They are true believers with no shortage of confidence, and they are fun to hang out with.  So, I'm a bit reluctant to offer this blog post because I am going to set forth my advice in writing--knowing that I might perhaps make future personal meetings redundant. (But I'm hoping they'll still call.)

To obtain the Holy Grail, you need to satisfy the following interrelated conditions:

1) A sophisticated data management system that, indeed, provides clinical advice that will be accurate in the vast majority of cases;
2) A plan to integrate that system into the various support systems that exist in a hospital so that it can be used in real time, i.e., as patient care is being delivered;
3) A plan to convince doctors and others to use the system;
4) A strategy for getting the procurement approved by the various high-ranking clinical and administrative officials in the hospital.

On the first point, what level of accuracy do you think is required to offer a decision support system that could have the confidence of doctors?  How would you test that accuracy?

On the second point, how long will it take to invent the interface between your system and the variety of clinical and administrative information systems that exist in your targeted hospital(s)?  Think about it this way:  How likely do you think it will be that you will get the time and attention of the CIO to install your system, as s/he is a bit busy with Meaningful Use projects?

On the third point, well, you know the issues.  Please don't think that because you've satisfied #1, above, that adoption by MDs will be assured.

On the final point, who within your targeted hospital(s) will carry the water for this project in the strategic and budgetary reviews with the CIO, CNO, CMO, CFO, and CEO?  Lots of people in a hospital can say "No."  Which of these people will say "Yes" and become your internal advocates?

If you can figure this all out and stay capitalized long enough to make sales and bring in revenue in a timely fashion, all will be well . . . if your approach truly offers a comparative advantage to the dozens of others trying to enter this arena.


John Moore ‏@john_chilmark said...

From Twitter:

2 points missed: Will it provide value at PoC? Don’t boil ocean - focus on a niche, deliver & grow

Dave said...

All good points though like real estate, 3 things matter the most -- evidence, evidence, evidence (of efficacy). Buyers & investors both want evidence so I'd find the most efficient path possible to evidence. I'd try to balance finding the most prestigious, yet smallest org that can provide credibility. My general advice to healthtech entrepreneurs is that health systems aren't the place to start. Expanding on your points, most health systems are mired in mega EHR implementations/MU/ICD-10 and have seemingly 100 people who can say no and few who'll put their neck on the line and champion something from a "risky" startup. Of course, there are exceptions but they are pretty rare in my experience. Progressively moving to bigger and bigger orgs is a good way to go. In fact, you can start the discussions with bigger orgs while the smaller/nimble orgs are executing. By the time you get further in the sales cycle with a bigger org, you'll hopefully have evidence from the smaller orgs.

Mitch said...

Excellent piece. I'd add a couple of points:

-They are often remarkably naive about the complexity of the medical system. They really don't get it. I could go on but I'm sure you have seen this. As a high ranking person at athena stated it, health care is "fiendishly complex." At least!

-They are arrogant. The general approach is: "I am from (Silicon Valley, MIT, Harvard, especially the Valley) and you in the health care world are slow, dim-witted, and we are here to save health care. Oh should it be so easy.....

Edward J Schloss MD ‏@EJSMD said...

From Twitter:

That’s definitely my sense, but pretty sure they’d call me the naive one. They seem unaware we’ve had Big Data in ICDs for yrs.

David States ‏@statesdj said...

From Twitter:

Still much to be learned in medicine & HC delivery. A little arrogance to challenge entrenched authority may be a good thing.

Barry Carol said...

It seems that there might be an opportunity for a wealthy research university with affiliated hospitals like Harvard or a huge hospital system like Ascension Health to form a venture capital subsidiary focused on healthcare and health IT. They could, presumably, not only provide capital but also plenty of doctors to offer a real world perspective about what they need and what might work. For data and information that doesn’t have to be accessed and acted on immediately, something like the Lexus-Nexis database that lawyers use might be useful.

Paul Levy said...

Certainly not at Harvard, Barry. That kind of cooperation is not part of the culture. Perhaps somewhere else . . .

Anonymous said...

This article linked below is apropo to your excellent post. Is health care really a service sector you can disaggregate silicon valley style? or an experience industry that turns away algorithmic solutions?

jvlbe said...

Interesting article, david
I would just like to add this viewpoint:
1° it is right: healthcare is a very complex ecosystem
2° on the other hand we need all means to get closer to the evidence, the holy grale. I'm sure that big data solutions might help to get closer to it.. by "5%", "10%"...but at what price? And in health care with closed budgets in many cases, we don't speak about price, but at what cost? So the real question when can we say that it is appropriate or even necessarily to use big data... and that evidence is totaly missing in the debate.