What AI-enhanced health care could look like in 5 years (VentureBeat)
This summer, KPCB partner Mary Meeker’s 2017 Internet Trends report singled out healthcare as a sector ripe with opportunity. The report proposed that the healthcare market, driven by a number of converging technologies, is approaching a “digital inflection point” and is currently positioned for rapid growth.
This is an understatement.
Due to increasing digitization of inputs since 2013, the amount of global healthcare data has been increasing 48 percent year-on-year, according to Meeker’s research. With the rising availability (and accompanying burden) of such rich informational resources, medical and healthcare practices are being reimagined on every front. Leveraging this explosion of data will create a complete transformation of both markets and service delivery norms. Data-consuming AI solutions will occupy a central role in this revolution.
Meeker’s analysis highlighted the opportunities surrounding digital innovation in patient empowerment and health management, improvements to clinical pathways and protocols, and preventative health. Innovation in each of these areas is inextricably tied to AI and machine-assisted processes. But what will AI-enhanced healthcare look like in practice?
Data-driven AI technologies are well suited to address chronic inefficiencies in health markets, potentially lowering costs by hundreds of billions of dollars, while simultaneously reducing the time burden on physicians. These technologies can be leveraged to capture the massive volume of data that describes a patient’s past and present state, project potential future states, analyze that data in real time, assist in reasoning about the best way to achieve patient and physician goals, and provide both patient and physician constant real-time support. Only AI can fulfill such a mission. There is no other solution.
Technologist and investor Vinod Khosla posited that 80 percent of what human physicians currently do will soon be done instead by technology, allowing physicians to focus their time on the really important elements of patient physician interaction. He wasn’t talking about the advent of robot doctors. He was referring to the manual data consumption and generation processes that currently consume so much of the physician’s time — and are much better suited to AI. Juniper Research analysis estimates that the use of chatbots in healthcare customer service alone can produce an “average time savings of just over 4 minutes per enquiry, equating to average cost savings in the range of $0.50-$0.70 per interaction.” Or consider the bothers of billing codification or data entry: These are time-eating, easily automatable tasks that neither justify nor require the skill level of the doctor. Such chores can and should be delegated to AI.

Healthcare in the age of AI

Within five years, the healthcare sector has the potential to undergo a complete metamorphosis courtesy of breakthrough AI technologies. Here are just a few examples:
1. Physicians will practice with AI virtual assistants (using, for example, software tools similar to Apple’s Siri, but specialized to the specific healthcare application). AI assistants will provide real-time and ongoing support and recommendations to the physician for diagnosis and treatment, as well as administrative support. A multibillion-dollar AI effort by many companies is already underway, notable in IBM’s Watson deployments or ventures such as Ada, which functions like a digital advice nurse. This is just the beginning.
AI virtual assistants will assimilate, analyze, and share massive quantities of data relating to individual patients in addition to entire patient populations. (A major part of the effort will first be aimed toward automation that maintains the privacy and security of individual patient data: The technology to do this is also advancing rapidly.) The data will be of all types — patient histories, emerging threat and epidemiology statistics, images, video, location data, physician comments and treatments, and much more — all at scales impossible to digest via human effort alone. Physicians will then use this knowledge to continue to improve treatment. Best practices will shift to require physicians to practice with AI virtual assistants.
2. Physicians with AI virtual assistants will be able to treat 5X – 10X as many patients with chronic illnesses as they do today, with better outcomes than in the past. According to a recent Research and Markets report, “A basic AI … today in clinical practices can be used for alerts & reminder, diagnostic, therapy planning, information retrieval, and image interpretation.” A tremendous amount of necessary labor can be automated and delivered concurrently via AI. Physician AI virtual assistants will have conversational ability and will be able to support the physician in messaging with patients, answering routine questions, suggesting treatment options, and more.
3. AI virtual assistants will support both patients and healthy individuals in health maintenance with ongoing and real-time intelligent advice. Our greatest opportunity for AI-enhancement in the sector is keeping people healthy, rather than waiting to treat them when they are sick. AI virtual assistants will be able to acquire deep knowledge of diet, exercise, medications, emotional and mental state, and more. Manually managing and recording daily nutrition, fitness efforts, etc. has always been tedious and difficult. But new technologies, including computer vision, natural language understanding, and machine learning, present interface capabilities that enable individuals to easily “show” or “talk to” their AI virtual assistant about what they’re doing. Further, AI virtual assistants can leverage other mechanisms to “know” what is transpiring via motion detection, IoT (internet of things) sensor input, and the like, to effortlessly collect valuable, personalized data. Patients will have a constant “friend” providing a digital health conscience to advise, support, and even encourage them to make healthy choices and pursue a healthy lifestyle.
4. Medical devices previously only available in hospitals will be available in the home, enabling much more precise and timely monitoring and leading to a healthier population. Already, a new generation of in-home health equipment is emerging, tightly aligned with advancing imaging and sensor technology for tracking biometric variables and collecting more frequent measurement data. For example, AliveCor’s mobile pad connects to smartphones and provides a personal EKG. This allows you to know anytime, anywhere if your heart rhythm is normal or if atrial fibrillation is detected. Another example is Scanadu’s test kit (under FDA consideration and soon to be launched), which uses AI to measure the level of chemicals in a urine sample and surface indications for several health conditions. There will be hundreds of other new examples in the very near term.
5. Affordable new tools for diagnosis and treatment of illnesses will emergebased on data collected from extant and widely adopted digital devices such as smartphones. Today’s smartphones already have high-resolution cameras, accelerometers, gyroscopes, touch ID, microphones, speakers … all of which facilitate health application. Small hand movements while holding the phone might provide a clue to Parkinson’s. A change in social network usage might indicate depression. Sentiment and/or vocal pattern analysis of speech might identify anxiety. And these are just a few examples of what can be analyzed via data already obtainable from an individual’s use of a smartphone.
6. Robotics and in-home AI systems will assist patients with independent living. Robots are leaving “the cage” where they’ve traditionally helped manufacture major products like cars, and now will be in your home, reminding you to take your medicine, assisting in everyday household tasks such as doing dishes and folding laundry, accessing out-of-reach objects for wheelchair-bound patients, helping the elderly into showers and baths, even increasingly providing emotional support, as with Japan’s robot pets.

The promise

You might think that AI is just going through a hype cycle that began over 50 years ago and led to disillusionment. I disagree. With the algorithm developments of deep learning, symbolic AI, computer vision, natural language, and machine learning combined with a smartphone — which puts the power of a supercomputer in everyone’s pocket and is always with you, always on, and always connected — we are at the beginning of the AI era. Multiply that power again by 1,000 with cloud computing, and there is no doubt we are on an AI curve of exponential growth.
Many of today’s familiar AI engines, personified in Siri, Cortana, Alexa, Google Assistant or any of the hundreds of “intelligent chatbots,” are still immature and their capabilities are highly limited. But don’t be misled — the best metaphor is that they are learning like humans learn and that they are in their infancy, just starting to crawl. Healthcare AI virtual assistants will soon be able to walk, and then run. Within the next few years they will be conversational, they will learn from the user, they will maintain context, and they will provide proactive assistance, just to name a few of their emerging capabilities.
And with these capabilities applied in the health sector, they will enable us to keep millions of citizens healthier, give physicians the support and time they need to practice, and save trillions of dollars in healthcare costs. Welcome to the age of AI.
Norman Winarsky is cofounder of Siri, Lecturer at Stanford Graduate School of Business, and advisor to Health2047 in venture development and commercialization. He is coauthor (with Henry Kressel) of If You Really Want to Change the World: A Guide to Creating, Building, and Sustaining Breakthrough Ventures (HBR Press, 2015).