I had the opportunity today to speak at the Open Science Summit at the Computer History Museum in Mountain View, CA; the meeting featured a number of interesting and unusually diverse talks, and many thoughtful questions. Like the annual Sage Commons Congress, this attracted an unusually involved (and articulate) group of participants.
My talk, entitled “Improved measurement: a path to better health for real patients,” aimed to present a concise, high-level overview of the key opportunities and challenges in the health space (nothing like modest ambitions). I’ll try to hit the high points here, many of which will be familiar to readers of this space; I would specifically highlight this four-part series on topics in healthcare innovation, and this recent post, co-written with MGH Physician-in-Chief Dennis Ausiello.
Goal: Better health for real people: health, not just disease; people, not molecules, cells, or rats; real, not only clinical study subjects (though I believe this is a critically important first step), but also real patients, in the real world.
"Without a functioning feedback loop, we are missing the essential opportunity to iteratively optimize and improve patient care."
Tool: Improved measurement – it’s difficult to improve what you can’t measure (see here, also here). However, can be two-edged sword – must avoid fetishization of metrics; many of the most important parameters are things that we can’t measure, and it’s important not to lose sight of or devalue these vitally important qualities.
Thesis: medicine’s greatest need is closing the feedback loop. Currently, patients are seen episodically (at best) by physicians then sent on their way. As discussed here and here, doctors have very little idea about what happens after patients leave the office, which is bad for both the patients (obviously), and for the doctors, who don’t have the opportunity to learn professionally and improve. Interestingly, Bill and Melinda Gates make similar points in the context of education in a recent WSJ essay. Without a functioning feedback loop, we are missing the essential opportunity to iteratively optimize and improve patient care.
Framework: As a product of BCG, I naturally conceptualize things via a 2×2 matrix; one axis = time horizon (short, long), the other axis = activity (research, application). Of course, real life is continuous not discrete, but the framework may be useful nonetheless.
On the long-horizon side of things – what can we do to generate truly breakthrough innovation? – I’d put my money on Eric Schadt-style systems biology, which I know is thirsty for more data, and would also hope to see the rebirth of human physiology as a more tractable – and hence more attractive –basic science.
On the short-horizon (where I think there is a particularly sizable opportunity to have significant impact), I argue that on the research side, our two greatest needs are improved technologies (obviously – and there is already a significant amount of effort in this space) and improved science, specifically “assessment science,” a term I first heard used by the FDA, who I believe has been ahead of the curve on this important issue (see discussion of SEALD towards the end of this piece). The point is that all measurement isn’t created equal, and it’s important to have confidence in the measures used, as well as ensuring that they capture the aspects of a clinical condition most important to patients. The goal of measurement remains not to maximize the volume of data collected, but to deliver relevant, actionable information.
Still on the short-horizon, there’s a tremendous challenge even after data is collected to ensure it is actually utilized. Virtually all physicians genuinely seek to deliver the best care possible, yet changing their behavior is notoriously difficult, as Peter Pronovost and others have discovered (discussed here and here). There’s a compelling case here to apply design thinking (defined provisionally as innovation that is human-centric, creative, iterative, and practical) to both how we think about patients (we need to get a much better understanding of their real lives so that we can do a better job of both assessing health and suggesting useful solutions) and also, and perhaps less obviously, how we think about physicians. Almost every doctor I know (and it’s a very high “N”) feel utterly pressed for time – most feel there already is far more to squeeze into a 15 or 30’ session than is reasonable or even possible. Collecting all the measurements in the world is not going to be useful unless it can be presented in an obviously useful fashion for physicians, and make their jobs easier (a lesson learned the hard way by many Silicon Valley initiatives in this space, discussed here). The idea isn’t to oversimplify – the complexity is real, and in some sense appropriate; rather, the challenge is to deal with the complexity on the back end of applications so that the user experience is as clean and accessible as possible (a concept I’ve borrowed, perhaps imperfectly, from Donald Norman).
These are the highlights, although there are several other benefits of improved measurement worth touching on as well, including:
- Identify new uses for existing drugs, the ultimate example of real-world phenotypic screening, which appreciates the value of a true, integrated read-out rather than a highly-reduced simplification (see this Boston Globe op-ed, co-written with Mathai Mammen, from earlier this year).
- Contribute to the identification and reliable quantification of real world needs, providing important opportunities for innovative medical products and protocols to demonstrate the value they are delivering (arguably the single most important way to improve the practice of medicine, especially if it includes approaches to assessing the associated economics).
- Enable sophisticated segmentation (especially if the measurement incorporates a range of phenotypic and environmental attributes), extending the thinking we now apply to genetic markers, and use this information to help physicians target different types of treatments to the patients most likely to benefit from that particular approach. This de-averaging could be extremely powerful – for example, it’s been very hard to show that classic disease management is profitable (discussed here); however, if you could select the patients most (and least) likely to benefit from a nurse phone call, for example, you could be more selective in your efforts, and use your limited budget more wisely.
Bottom line: improved patient-focused measurement, while not perfect, could be profoundly enabling in the short term as well as beyond– but capturing this potential will require thoughtful science, involved patients, and inquisitive physicians, as well as the shared commitment to iterate and optimize around the common goal of improved health for real people.
David Shaywitz, M.D., is an adjunct scholar at the AEI.