The World Health Organization reports that 45% of member states have less than one physician per 1,000 population. With the Kaiser Family Foundation reporting 27 million Americans lacking health coverage in 2017, access to healthcare is not just an issue in the developing world. These numbers help highlight healthcare’s two idealized goals:
- That healthcare be accessible to everyone; it’s a human right.
- That we all die in our sleep at 115 having incurred $0 lifetime healthcare cost.
We will forever be driving toward these two asymptotic goals. Achieving the former requires healthcare that is scalable and frictionless, while achieving the latter requires healthcare that is predictive and preventive. Healthcare is also a zero-sum game. Every dollar spent in one area is a dollar that cannot be spent elsewhere. Consequently, healthcare dollars must be prioritized with maximum impact in mind.
Doctors serve as the gateways to healthcare today, with drugs and devices generally serving as the means of therapy. Because physical “contact” with the system is required in all instances, healthcare’s scalability is limited. And while effective, this canonical approach to healthcare is often reactive, not predictive.
Artificial intelligence brings the next great breakthrough in healthcare, promising to help deliver on healthcare’s two primary goals. AI today tends to be classified by technology type: convolutional neural networks, support vector machines, generative adversarial networks, etc. But it might be better classified simply by human impact: contact vs. non-contact.
As with doctors, drugs, and devices, “contact” AI requires physical contact with the healthcare system. Into this category of AI fall the many machine learning techniques to classify pathology images or aid in drug discovery. All worthy solutions to be sure, but none can be deployed without the patient’s physical contact with the healthcare system. A more scalable and arguably larger impact for AI lies in “non-contact” solutions: combining AI with the endless data signals from mobile devices to predict, prevent, and sometimes even treat health issues. Because data is collected, analyzed, and acted upon remotely, non-contact AI helps deliver predictive and scalable healthcare without the need for physical contact with the patient.
Non-contact AI built upon data signals from mobile devices has been developed for a range of healthcare applications already. A prime example has been AI combined with smartphone cameras for diagnosis in teledermatology. Non-contact AI for diagnosis, prevention or treatment has also been delivered in areas that include diabetes, cardiac arrhythmia, coronary artery disease, PTSD, cognitive decline, Parkinson’s, depression, various genetic disorders, pancreatic adenocarcinoma, drug-to-drug interaction, and population health. Fully developed, the potential impact of non-contact AI solutions in healthcare is enormous. Not surprisingly, both Google and Amazon have patented non-contact AI technologies to measure health using mobile devices.
Healthcare is a human right, but we have struggled to make it accessible. Furthermore, healthcare’s zero-sum reality requires prioritizing our resources. Even with these challenges, we are at an exciting crossroads. Mobile devices are multiplying across human populations, deterministic data signals are exploding, and (contact and non-contact) AI presents a transformative new weapon in healthcare’s arsenal.
The predictive power and scalability of non-contact AI yields the broadest human impact, and perhaps merits healthcare’s greatest focus; it may yield our greatest payoff as technologists and entrepreneurs. Uniquely, the human impact of non-contact AI may both win a Nobel Prize and also fuel tech’s next IPO.