Managing your health data: no easy task
Information on more and more aspects of our health is increasingly available: just take a look at the possibilities offered by a simple iPhone for monitoring vital signs, activity data of all types, along with devices with which you can exchange information on a regular basis. More and more devices are competing for a niche in a market that takes in fitness to specific controls to deal with all types of ailments. Currently, we can monitor our heart rate in just about any situation, from baseline to exercise, maintain blood glucose control or even analyze urine, as well as being able to relate them with data from our genome and treat them with the appropriate therapies.
The problem with all this is processing the data: the apps we use to collect them will usually allow us to keep an orderly record of them, but will not venture any diagnosis beyond a simple indication, because the diagnosis is generally considered to be reserved for the work of a doctor.
But most public and private healthcare professionals would not be too impressed if their patients were to turn up for an appointment loaded with countless data in their respective apps: most medics are too busy face the task of examining them, while others would simply dismiss the information as irrelevant because it comes from devices that are not approved for medical practice. Thus, the infinity of data generated in our daily monitoring that could help with preventative health care goes to waste. The result is the main problem of health systems, beyond their cost: we only connect with them when we are already sick.
How to improve things? Given that we have more and more input data and plenty of studies, it seems reasonable to think of this as a job for machine learning: use a set of algorithms to evaluate the parameters supplied by our sensors and that generate a first diagnosis that would mean only potentially serious cases would be passed on to a physician.
The patient, in such cases, would enter the health system with a pre-diagnosis and a set of readings that would be repeated to rule out a poor reading or sensor malfunction, and be appropriately valued. Such a system would not only allow for many ailments to be dealt with before they reach levels of concern, but would also allow hospitals to obtain more data in comparable situations that would enable progress in pre-diagnosis and potentially advance the development of medical science as a whole, as some major technology companies seem to be pushing for.
Who in the health system will be the first to undertake such an approach requiring sensitive data treatment and specific machine learning development? Private health insurers could give it a go, but they could find themselves mistrusted by some patients concerned that troubling data could send their policies through the roof.
At the same time, it seems clear that there is a market for such preventive health monitoring, and perhaps this is just the way to go about it.
Can mass data generation become the key to keeping us away from the doctor for longer and only seeing one when we really need one, or even make decisions when there’s still plenty of time to treat diseases, with a preventive mentality? Could wearables, sensors and machine learning provide the means for more cost-effective healthcare systems?
(En español, aquí)