AI and machine learning is opening up possible new avenues in disease detection — but just because we can do something, does that mean we should?
Last month, I wrote an article called “3 Amazing Ways Google Search Data is Improving Healthcare,” that discussed the notion of using search engine data to diagnose illness before patients are even aware that they might be sick.
I recently came across a Wired article by Dr. Sam Volchenboum, the Director of the Center for Research Informatics at the University of Chicago, and a co-founder of Litmus Health, a data science provider for early-stage clinical trials, that explored this idea in depth. Here are a few of the key takeaways from his piece.
From a data science perspective, says Dr. Volchenboum, the world is effectively becoming “one big clinical trial.” Internet search, social media, mobile devices, wearables, etc. are generating a steady — and staggeringly large — stream of information that “can provide insights into a person’s health and well-being.”
We’re not quite there yet, but it’s entirely possible that in the very near future, platforms like Facebook and Google will be able to alert someone to the possible presence of a disease before they’re even aware of it. While, in theory, this kind of technology would have the potential to save lives, Dr. Volchenboum aptly points out that when it comes to electronic patient health data, it’s never black and white.
In order to create a predictive model, a platform like Facebook would have to start by working backwards. Dr. Volchenboum explains, it would generate “a data set consisting of social media posts from tens of thousands of people will likely chronicle the journey that some had on their way to a diagnosis of cancer, depression, or inflammatory bowel disease.”
Then, using machine-learning technologies, a researcher or provider could analyze all of those disparate data points, taking into account the “language, style, and content of those posts both before and after the diagnosis.” This would allow them to create models capable of identifying similar behavior, which, in theory, would suggest a similar outcome down the road.
While such “early warning systems” are not yet in place, the underlying technology necessary to develop them certainly exists — the advanced predictive and machine-learning algorithms powering Facebook and Google’s advertising platforms basically use the same concept, but simply employ them to different ends.
I agree with Dr. Volchenboum that yes, we should start leveraging the vast amounts of consumer data in ways that benefit society as a whole, but that we also need to be very careful if and when we attempt to do so.
As we all know, the companies behind today’s biggest digital platforms detail how they plan to use consumer data in their terms of service; but as we also all know, few people actually take the time to read the terms of service. So, while these companies may be covered from a legal perspective, they’re not actually providing a functional window for patients who may be concerned about where their data ends up.
If this is the path we ultimately go down (and I’m quite sure it will be), we need to make sure it’s a highly transparent, opt-in system for those patients interested in participating. That means spelling it all out in terms that patients can actually understand, ensuring their data remains protected, and, if they choose not to participate, respecting that decision and keeping their data private. As patients continue to take a more active role in their health and treatment decisions, it’s likely that many would be in favor of this kind of technology — we just need to make sure it’s built upon a foundation of trust and respect.