Dr. Alexa Will See You Now: Can We Trust Digital Assistants With Our Health Data?
It’s hard to imagine a field where data is more valuable than in healthcare. Monitoring and interpreting crucial health indices can make the difference between an early diagnosis and a late one—and that, in turn, can be the difference between life and death for some people.
As neural networks and machine learning algorithms get better at interpreting these signals, and even diagnosing certain illnesses, there’s a clear drive towards creating a fully-integrated medical assistant: one that can converse with you about your health while simultaneously using data from wearable devices, blood-sugar measurements, and possibly more complicated tests to diagnose any conditions.
In an ideal world, this would flag any potential problems early, allowing patients to get the best care—and, when the patients see their doctor, the healthcare staff would have access to a much better record of their recent medical history to assess the best course of treatment. Once treatment was ongoing, the medical assistant could monitor its progress, analyze its effectiveness, and update the hospital.
If properly anonymized, this could provide a wealth of additional data allowing medical professionals to improve treatments and give patients better information. The result, in this optimistic vision of the future, is better, more efficient healthcare for patients.
HIPAA-Drive
But healthcare is also an industry, especially in the US, and it can be immensely profitable. There will always be a tension at the blurry borderlines between data collection for the purpose of improving a service and surveillance capitalism. A recent example of this is Amazon’s announcement that its Alexa voice assistant and associated devices can handle customers’ healthcare data.
Specifically, the data-handling protocols for Alexa are now compliant with regulations under the US Health Insurance Portability and Accountability Act (HIPAA), which sets the standards for storage, transmission, and use of sensitive healthcare data for patients.
Amazon also announced partnerships with six other HIPAA-compliant companies, which are building Alexa apps (“skills”), allowing Alexa owners to check the status of the shipments for prescription medications or book emergency appointments with certain healthcare providers.
These may seem like small steps—is it really that different to booking appointments over the phone, or via another app?—but they represent part of a long-term strategy by Amazon to move into the healthcare industry.
Customers can even monitor and analyze their blood sugar results via Alexa, communicating with wireless glucose monitors. Amazon sees a profitable niche in helping patients with diabetes manage their condition, and it’s even more profitable if you can also sell them the insulin, a global market estimated to run into the billions of dollars by itself.
Amazon’s Strategy
Last year, Amazon bought Pillpack, an online pharmacy company that ships medications directly to users, for $1bn, aiming to do to pharmacy what it has already done to many brick-and-mortar bookshops and electronics outlets. Earlier this week, eagle-eyed reporters spotted a sign-up page for Pillpack on Amazon’s website, suggesting that the service is ready to be rolled out shortly. It seems inevitable that this will be integrated with new Alexa skills and data management, to encourage customers to buy their pharmaceuticals via an Amazon-owned company.
Moving into healthcare has been a long-term goal for Amazon for a while. Last year, it was reported that Amazon’s healthcare and wellness group had made some notable hires from other healthcare companies, and was seeking to make its device HIPAA-compliant, allowing Alexa to move towards becoming a more powerful digital health assistant. Meanwhile, the company already has much of the delivery infrastructure in place to become a major supplier of medicines to individuals and hospitals.
Now, with HIPAA compliance, Amazon can use Alexa to legally share the healthcare data with medical professionals, patients, and the third-party apps it forms partnerships with. If managed well, this will undoubtedly provide benefits to consumers. For less serious conditions, you could imagine the entire process, from diagnosis and testing, to medication, to monitoring and follow-up, being done entirely via Alexa and Amazon’s delivery services.
An Internet of Spies?
But of course, this comes with trade-offs. One potential customer for Amazon’s healthcare databases will be insurance companies, looking to fill in incomplete databases so that they can more accurately price their products to individuals, swelling the profit margins for the industry. In a world where Facebook knows that you’re pregnant before you do, the price of your health insurance could be dictated by a black-box neural-network score that’s analyzing endless streams of data: what you eat, what you buy, what you search for online. Adding Alexa, or connected healthcare diagnostic devices, could just as easily allow your insurance company to know you’re sick before you do.
Given that Amazon has patented a skill that allows Alexa to pick up on coughs—and, potentially, to distinguish between the flu and something more serious—you can expect to be swamped with adverts for cough drops and cold medicine if you ever happen to sniffle next to your Echo device.
Perhaps more personal, and more worrying, is that similar software will be used to analyze your emotional state. This was considered as part of the diabetes management plan back in 2017. Living with the condition can be stressful, and that stress can in turn affect your blood-sugar levels.
An integrated system that monitors your blood sugar, reminds you to take your insulin, calculates the appropriate dose, and even orders more when you’re running low might be a boon to patients. Such systems do already exist and are widely used without passing data to Amazon.
But if that same device starts analyzing your mood via your voice and blood sugar, and uses that to target advertising or build up a profile of you that is then sold to other companies, it becomes a gross invasion of privacy under the guise of providing better care. And, ultimately, if the algorithms that make decisions about the healthcare you receive are impenetrable black boxes that seek to optimize profits for private companies, as well as healthcare outcomes for individuals, the utopia we started this article with looks very different indeed.
Balancing Benefits With Privacy
In the digital economy, we all face daily choices and daily trade-offs of this kind: sacrificing privacy for convenience. Often, though, we aren’t given a choice: a necessary condition of using a service is that it collects data on you and builds up one of countless profiles of you that exist in databases around the world. HIPAA theoretically gives patients the right to access the healthcare data that is stored about them, and to limit access to that data.
It’s a start, but it still feels that individuals are deliberately kept in the dark about algorithmic influences on our lives, and increasingly face decisions that are made by processes they cannot comprehend, using data they had no idea was being collected or analyzed in this way, and which cannot be easily queried or reversed.
There are many things we would like the future of healthcare to be, but Kafka-esque is not one of them. The balance between the many potential benefits of big data analysis in healthcare and the potential for exploitation and privacy violations surrounding that data should be a hotly-debated topic for years to come.
