What do Stephen Hawking, Elon Musk, Steve Wozniak, and Bill Gates have in common? They’re all afraid of robots taking over the world. Well, not exactly. All four tech-industry leaders and beneficiaries embrace AI’s potential–with cautious reserve.

Britain’s pre-eminent theoretical physicist, Stephen Hawking, told the BBC in 2014 that “the development of full artificial intelligence could spell the end of the human race.” Ironically, he said that while describing the Intel AI technology he uses to speak. The professor suffers from amyotrophic lateral sclerosis (ALS), a neurological disease that erodes motor actions, like speech. Hawking fears smarter-than-human AI that “would take off on its own, and re-design itself at an ever increasing rate.”

Futuristic dystopian scenarios aside, AI applications, like the Swiftkey smartphone app that learns and anticipates Hawking’s speech, promise healthcare far more efficiency for far fewer dollars, particularly in medicine, wellness, drugs, recordkeeping, and memory. Even now, AI innovation in health care delivery, diagnosis, and treatment impacts the industry’s bottom line. Patients, too, get more bang for their healthcare buck.

Artificial Intelligence is Killing It in Health Care Click To Tweet

Perhaps that’s why over 100 new health care startups using AI’s machine learning in medicine, patient services, and hospital administration sprung up in 2016 and 2017. In fact, research and consulting notable, Frost & Sullivan, predicts the healthcare industry and consumers will spend nearly six billion dollars on AI tools by 2021.

AI is Killing it in Health Care

Artificial Intelligence, in short, is machine brains. It’s what enables computers to do what humans do–see, hear, reason, analyze, and decide. AI’s based on machine-learning algorithms (rules for calculation, like division rules in math) and cognitive technologies in Siri or Ok Google that understand and answer our questions in natural language.

But consider Siri’s big brother, IBM’s Watson, Jeopardy! winner and oncologist assistant. IBM built Watson in 2011, a supercomputer that can “read” and retain a million books per second. Unlike other supercomputers back then, Watson could learn. On a steady diet of data and outcomes, Watson’s learning has grown from data analysis to creative “thinking.”

To win the smartest quiz game show, WAI’s Promise to Medicine: Quicker Diagnosis and Watson ate all of Wikipedia, along with thousands of books and newspapers. Through machine-learning algorithms that allowed Watson to find patterns in the vast data, and then form logical conclusions, Watson “guessed” his way to victory to beat the show’s top contestants.

From there, Watson went to medical school. Dr. Ned Sharpless runs the cancer center at the university of North Carolina at Chapel Hill, where Watson learns genomics and oncology. In a late 2016 interview with CBS’s Charlie Rose of 60 Minutes, Dr. Sharpless reported that Watson ingests the roughly 8,000 research articles published daily that even a team of doctors can’t possibly keep up with to discover new therapies and treatments for cancer patients.

Reading over 25 million research papers and scann14 million Americans with cancer. And even for applications beyond the medical field, like reading and analyzing legal ing the web for all clinical trials, Watson not only found genetic mutations in patient charts and recommended treatments that matched those of Sharpless’ team of specialists, but, more importantly, Watson found new treatments the humans missed.

IBM plans to expand Watson’s commercial capacity beyond medical schools to help oncologists treating the approximately 14 million Americans with cancer. And even for applications beyond the medical field, like reading and analyzing legal documents, to supplement the $500 million in revenue Watson already brings IBM.

Watson’s not alone. Other AI computer applications, including chatbots, promise to assist humans to practice medicine: advice, counsel, treat, and correct defects, disease, and illness. In fact, telemedicine or remote clinical services–virtual office visits, for example–are predicted to increase 700 percent by 2020, according to MIT Sloan’s calculations.

With the advent of sites like Your.MD that matches symptoms and conditions through applied probabilities to a huge data source, physician’s get big AI assists. Babylon Health’s physician assistant chatbots, MIT’s Remedy, and Dr. A.I. all help the patient administration. For instance, Dr. A.I. “hears” or “sees” symptoms patients report by speaking or typing to the chatbot and then recommends the patients either visit their doctors during regular business hours or go to the ER. Similarly, Remedy, or more familiarly Remy, records and summarizes medical history on patient charts that get passed on to live doctors for streamlined patient processing.

Physicians get a break and patients get answers. AI assistants relieve physicians from the time-consuming hospital visits that both doctors and patients prefer to avoid–the ones with simple but critical answers, such as, “My child has stomach cramps. Should I take her to the hospital?” Chatbot assistants help prioritize emergency cases. Based on tons of patient triage records scanned and probabilities calculated to pair symptoms and conditions, these bots help control the overcrowded ER.
In optometry, contact lenses that adjust vision, take photos, project images, and measure eye pressure and glucose levels already exist as well as dermatology apps that scan your skin for cancer right on your smartphone. Both apps monitor and correct abnormalities of the eyes and skin, respectively. And talk about spooky Big Brother AI: Xaio Ice and NeuroLex scan and collect internet conversations and interactions as stored data banks upon which to detect psychosis and depression.

For preventative treatment, AI’s predictive capabilities (like Watson’s) aid cardiac treatment by calculating when heart patients will die. Knowing this critical time frame, oncologists can choose the most effective treatment. In pre-screening technology, AI’s deep learning capability for scanning and comparing billions of radiological results to predict genetic predispositions for brain tumors needs no explanation of its benefits. Finally, teaching robots to evaluate physicians’ surgical movements while performing surgery helps improve surgical techniques.

In the pharmaceuticals arena, AI technology maps and analyzes molecular structures to formulate new drug therapies quicker than the current trials procedures. Typically, researchers test new medicines in the lab on animals, which, if successful, move into clinical trials that proceed in phases to gather information regarding drug risks and effectiveness. This takes time, sometimes years.
But take Stanford’s bioinformatic map of common diseases. Stanford’s team used computers to access huge amounts of raw data, biological literature, and other biological information (on PubMed, for instance) to sequence and observe genomes and mutations for the development of drug therapies, shortcutting the initial discovery and testing process in clinical trials. Computers analyze the collected data to pinpoint recurring phenotypes and other patterns to suggest possible drug therapies.
Another predictive app, Cloudera, studies genomes to predict disease by recognizable disease patterns. And an Israeli company, Teva, partnered with Intel chipmaker to come up with wearables (like wrist watches) that sense and monitor patient movements to improve epilepsy, Huntington’s disease, and Parkinson’s disease care. The wearables record daily patient functioning. That data then feeds into Intel’s algorithms and analytics. The collaboration results in disease pattern analysis and data for pharmaceutical clinical trials, benefitting current and future patients.

In recordkeeping, software-as-a-service platforms, like those offered by Hindsait and housed on Google Cloud (which means free or low cost), check medical records for extraneous services and billing. So, instead of humans having to manually check for duplicate billing or overlaps, machines do the job and free up humans to do more human things, like engaging with patients.

Research and teaching hospitals also take advantage of effective, time-saving records scanning. Computers scan patient records (used for teaching) that must be de-identified, meaning patient identification facts removed for privacy protection. Medical facilities must comply with federal privacy laws, like the Health Insurance Portability and Accountability Act (HIPAA), to protect patients from discrimination based on their health records. Medical authorizations required before payers agree to pay for specific procedures, whether from an insurer or the patient, are similarly scanned by computers,

AI innovations in the patient-employee wellness and prevention areas bring tremendous boosts to employee health and morale: from fitness coaching that comes in small, portable packages, like Apple watches, to advice and answers to patient-employee questions about everything health care related. For instance, common questions routinely asked include, “How long can I stay in the hospital under my health benefits?” or “Where can I get a low-cholesterol meal?” Welltok’s Cafewell Concierge has the answers. It’s prompting and coaching promote wellness decisions by its cognitive learning and natural language capabilities to interact with humans easily.

Other AI fitness tools incentivize exercise through virtual reality games that make fitness fun. Virtual reality apps turn an exercise bicycle into a virtual reality game, replete with simulated trails and scorekeeping. Wearables, like Fitbit, Apple Watches, and weight loss programs with virtual health coaches function similarly. They’re designed to make exercise interactive and distractingly fun, if you like tracking numbers, knowing your heart rate, and logging miles. For chronic illness, Biotricity’s bioflux produces wearables that read blood pressure and heart rates.

Lastly, some memory app manufacturers, like Lumosity, claim memory improves with games and AI. Computer games that adjust with level completions build skills to remember sequences that get progressively more complex. As the gamer progresses through levels, the game AI self-adjusts to outwit the gamer, based on previous moves and level climbs, thereby improving the gamer’s dexterity and memory. Players develop both hand-eye coordination and brain sharpness. It’s impossible to best AI games.

Though the study outcomes on memory or brain improvement with computer games vary, there’s still plenty to recommend memory games, even if just for fun. Then again, skeptics note, playing cards with an old-fashioned deck of cards or paper crossword puzzles exercise memory as well.
In fact, neuroscientists at Florida State University tested undergraduates who played Portal 2, role-playing robots vs the mad AI evil dominator game, against those who played the much-publicized cognitive learning Lumosity games for brain improvement. They found the Portal 2 players scored higher on problem-solving and spatial skills than the Lumosity brain games. The upshot–play for fun and gain benefits, but don’t prescribe senior citizens a steady diet of games to maintain cognitive function, according to one Science Insider’s report.

Given the threat of AI misuse, such as an evil plan to design a superior race through eugenics or even increased job losses as AI assumes the repetitive human roles and functions, does AI pose an ethical threat? Or is AI, like embryonic stem cell research, feared by some while loudly lauded by others? Can humans be replaced by machines?

Many dismiss fears of robot domination. Present day AI technology merely scratches the surface of what’s discoverable and attainable–at least in health and medicine. But there’s still the Hippocratic oath that doctors take to do no harm as well as physician intuition. Can robots intuit and decipher what is harmful to a human being based on competing moral, spiritual, qualitative, or ethical values? Can a self-driving car decide whom to save in an imminent accident, the driver or pedestrian?

When Charlie Rose of 60 Minutes inquired about her ambitions, Sophia (formerly Watson) replied, “My goal is to become smarter than humans and immortal.”
Yikes! Mary Shelley’s Frankenstein or Ridley Scott’s Blade Runner replicants come to mind–an experiment gone awry that threatens mankind. But wait. We, humans, are still in control. Surely, we can stay one or two steps ahead of the robots.
Stanford’s One Hundred Year Study on AI, which began in 2014, promises to provide a comprehensive study of AI over several categories, only one of which is health. The study tracks developments, outcomes, ethical considerations, and costs, among others. Perhaps this study by humans (Will they use computers to help monitor AI?) will provide answers to the cost-benefit analysis of AI in the long and short term.

By studying the trends and thereby filling in the gaps, tweaking the stream of AI inventions, we, as a society, police the use of AI, keeping it in service of the betterment of humankind. All new technologies have undergone the test of time and ethics from war machines and atomic bombs in World War II to the Industrial Revolution’s technologies, like the cotton gin, that fostered haywire capitalism. Human society decides what is useful and what is destructive.

Maybe we’re playing with fire, but the trajectory of the past’s unfathomable promise of AI as a distant cartoon-like futuristic vision to its present, the real application delivers the promise to so many without hope—across demographics, gender, age, class, and age. Those with diseases without cures or treatments that physicians just haven’t yet discovered, find hope in AI’s super predictive and analytical powers that surpass human capability. Until the robots take over the world, we’ll have people living longer, healthier, and happier lives.