Here’s a development I considered outright ground-breaking: a Calgary-based law firm is using its client’s Fitbit data to back up her personal injury case in court. The point is to prove that the person’s activity levels have been negatively affected by an injury she suffered in an accident four years ago. The used data isn’t raw data from the Fitbit band, but it’s first crunched by Vivametrica, which matches it against that of the general population.
Based also in Calgary, Vivametrica is a startup that has developed an analytic platform designed specifically to deal with health and wellness data collected by wearable devices. I got to meet a couple of folks from the team in London last month, and they described their approach as being a health company working on data, rather than a data company working on health. To me, that’s an important distinction to make, considering how critical domain expertise often is in order to make sense of data that comes from “physical-first”, IoT-style environments. Analyzing data from e.g. vehicles or infrastructure, let alone humans, is more complex than doing the same to data from digital-first domains like online behaviour or financial transactions.
It may be difficult, but there’s no denying that it’s transformative as well. As we at ABI Research like to put it (for example, in our recent article for IEEE’s new IoT newsletter), these emergent forms of data analytics will shed light on assets and processes that have traditionally been opaque to analysis. That way, solving physical-first problems becomes less reliant on subjectivity and anecdotal evidence.
In this regard, the court case in Canada is a significant one, as it involves a “problem” that is typically solved through expert opinion, which no matter how well-informed is in many cases inherently subjective. Besides the law firms that handle injury cases, the outcome will no doubt interest also health insurers facing large claims. Similarly, health-care providers, and especially the ones that are under pressure to prioritize expensive operations and treatments, could use data from medical wearables to evaluate who would benefit from them the most.
Many people may find the prospect of all this somewhat contentious, since it essentially means that many life-changing decisions may soon be based on the sort of sensor data that didn’t even exist until very recently. But on the balance of things, it’s hard to argue against the benefits of a more data-driven approach. I don’t think it’s realistic to expect that such decisions will ever be entirely outsourced to algorithms without any checks and balances; the opinions of human experts will still be needed to interpret and evaluate the findings. And an expert’s opinion, at the end of the day, is surely more valuable if it takes advantage of relevant and adequate data than if it doesn't.
That is not to say that some possible uses of health analytics don’t warrant debating, per se. Quite the contrary. For instance, take the underlying health-insurance paradigm, which essentially assumes that the insurers cannot quantify the risk of each customer in full, for if they could then no insurer would cover anyone with an unprofitably high level of risk. Now, what will happen if that risk – some of which is beyond the involved person's own control – becomes someday fully quantifiable? It’s a big question, and nobody has the answer.