How Mobile AI Will Transform Global Health
The Invisible Revolution Continues
In the last 15 years or so, the mobile phone has exploded across the developing world. The benefit was initially just one of distance communication — telephony and SMS. As handsets get cheaper and more capable, and network speed improves, however, it’s clear that these devices haven’t just “leapfrogged over landlines” — but laptops as well: the mobile phone revolution is clearly a mobile computing revolution in progress (a point I noted 9 years ago in a BBC article) — and mobile AI (artificial intelligence) will be the next stage in this revolution.
The benefits of mobile phones to global health have already been enormous, starting with the incalculable value of easy and inexpensive access to communications. Those who worked in the field prior to mobile phones can easily verify that every single field activity, every single management task, was more difficult when landline phones were the only game in town, and those were luxury items. And for citizens of poor countries, the mobile phone may be the greatest labor saving device ever created, allowing mothers, for example, to call clinics or pharmacies for advice or information rather than burning scarce calories walking.
Now, as computers, that value continues: access to weather reports, market prices, news — indeed, nearly the whole of human knowledge via sites like Wikipedia — along with incredibly useful apps that help individuals (for example by providing basic money transfer and banking) and organizations (with mobile data collection and group communications).
But we are beginning to realize that all the benefits up to now have only been prelude to something with even greater impact on international health: the rise of artificial intelligence, delivered to even the poorest people in the world via the mobile phone.
AI: Using Algorithms to Do What Doctors Do
When we talk about artificial intelligence, we mean the idea of using computers to perform activities that mimic human thought and produce similar results. Some examples include image recognition, where in just a few years the field has advanced from being unable to tell a cat from a dog to being able to tell one dog breed from another — and more importantly, to being able to examine skin photographs and diagnose skin cancer more accurately than expert dermatologists, as detailed in both an original article in the Journal Nature, and in a writeup in the New Yorker:
Of course, researchers are not simply applying this to dermatology, or even just to image recognition (though initial studies with interpreting medical images are just as promising): what the AI is being used to do is to recognize patterns. And patterns of sound, patterns of sleep, even patterns of behavior are just as amenable to this type of approach as are patterns of skin photos and chest x-rays.
Other examples include:
ResApp Health, who have used AI to develop an application that listens to the sound of coughing and breathing and can accurately then diagnose pneumonia or asthma.
IBM’s Watson, which can diagnose cancer, and select appropriate treatment, more accurately than expert cancer specialists.
AiCure, a mobile app that uses AI to verify medication compliance (it can watch you ingest your meds) — and which could be used to scale directly-observed therapy (DOT) for TB (currently dependent on community-health workers) at low cost to all the places that current cannot afford it.
NIH facial image recognition algorithms that can diagnose genetic disease using a smartphone camera.
The type of AI described above, used to provide a service that previously required highly-trained and scarce (and thereby very expensive humans) by means of an algorithm that requires no food, water, or salary, could easily cause havoc within the community of physicians, by causing the price of medical expertise to plummet.
In rich countries, this will affect the number of doctors that need to be trained (or employed) to meet health needs.
In poor countries, where there have never been anywhere near the number of doctors required to meet health needs, the impact would be far greater: it would bring medical expertise to literally billions of people with smartphones who have never before had access to any kind of such expertise.
In the same way that the mobile phone allowed poor country citizens to leapfrog over landlines, and then over desktops and laptops, it seems it is poised now to help them leapfrog over doctors — with the end result of better health for all.
Originally published at Magpi.