In the development of drugs, it's too difficult to change things on the fly, and to rapidly pivot in response to an appreciation of customer need
A few months ago, I heard a young design entrepreneur named Aza Raskin talk about his idea for a consumer health company, MassiveHealth, built around the concept of providing rapid feedback. For example, if you had a skin dye that faded a certain amount each time you took a dose of your antibiotic, you would be more likely to complete the full course.
Skip ahead not very far. Recently, MassiveHealth launched its first, free app (dubbed an experiment), called the Eatery. The idea is that you take a picture of your meal and rate its healthiness, which is then shared with other users. You benefit, as I understand it, by thinking more about your food and by getting input on your food from other users. What the company itself gets is not yet clear. They've shared some pretty maps of San Francisco and New York City showing where people are eating more vs. less healthy foods, and they've drawn some fairly general conclusions about how the supposed healthiness of our food changes during the day (good at breakfast, bad during the day, partial recovery at dinner).
At least as important, I'd imagine, they have an engaged group of users who seem (at least at this early stage) to be interested in interacting with the platform, and thus contributing to the development of the emerging data set; after only a week, more than one million food ratings were reportedly received.
In the current biopharma model, researchers spend years developing a product without any idea of whether it will work or be accepted
As I've followed the evolution of MassiveHealth, I've been struck by some of the profound differences between a tech start-up (even one ostensibly in the healthcare space) and a biopharma start-up. In the standard biopharma model, you spend years developing a product, without having any real idea of (a) whether it will work, (b) whether it will be safe and well-tolerated, and (c) whether by the time you've demonstrated (a) and (b), anyone will care, or pay you for your efforts. When you develop a new drug, most of the relevant properties of the product are pretty much baked in at a fairly early stage; you can tweak the formulation a little bit (to make it longer-lasting, say), but otherwise, you have what you have, and the challenge is figuring out just what this something is, and determining who might benefit most from it. Many have compared the process to an unforgiving lottery: Make a mistake at any point (choose the wrong indication, the wrong study design, or the wrong study sites) and you're out of the game; execute flawlessly and you buy yourself only a chance to see whether or not your number is drawn.
Unfortunately, getting to the drawing takes a lot of time and money: most of the cost of drug development (which generally exceeds $100 million for an individual program -- and this figure doesn't include the cost of all the failures) isn't from coming up with the particular molecule, but rather in putting it through the increasingly expensive series of clinical studies required to see if it's actually going to work. Of course, once you've successfully run this gauntlet, not only do you have something that's demonstrably effective (at least in the setting of clinical trials -- see here), but there's also a pretty high barrier for potential competitors, at least until your patents expire.
In contrast, MassiveHealth has managed to get a product in customer hands after a few months of work. True, they've probably not made any money thus far, and (not insignificantly) it's entirely unclear whether they have, or will ever impact anyone's health, massively or otherwise. Nevertheless, they have an extraordinarily powerful opportunity, at this very early stage, and after spending (I suspect) very little money, to learn, gather feedback, iterate, and explore: In an organismic sense, they can carefully assess their environment and respond adaptively -- evolve their product based on demonstrated customer needs. And they can do this very rapidly so that even if just one aspect of their platform is interesting, they can rapidly pivot and exploit it (similar to the way Twitter developed).
The ability to make cost-effective exploratory efforts is a powerful enabler of innovation, as Peter Sims has highlighted in Little Bets (my Wall Street Journal review here, implications for pharma discussed here and here), and the need for successful start-ups to rapidly collect data and adjust course is more the rule than the exception, as John Mullins and Randy Komisar thoughtfully discuss in Getting to Plan B (listen to this interesting podcast of Komisar at Stanford).
Unfortunately, drug development is far less conducive to this sort of exploration; as I've discussed elsewhere, the cycle times tend to be far too long, and the costs are way too high. As a result, it's a lot more difficult to change things on the fly, and to rapidly pivot in response to a new appreciation of customer need.
Not surprisingly, there's been a lot of interest in ways to streamline the drug development process. One especially attractive way is to identify new uses for existing drugs (as these have already been carefully vetted), an approach I've previously argued might be especially amenable to a crowdsourcing model. More detailed patient segmentation -- more comprehensive phenotyping -- could also be helpful, as many clinical studies could be smaller and in many cases shorter if you could more precisely target your intervention to the patients more likely to benefit, and thus boost your effect size. The development of new, highly predictive models, including in silico, preclinical (i.e. animal), and especially experimental measurements in healthy human volunteers, would be especially valuable, and would allow for more rapid iteration and optimization.