by Dr. Rob Nelson –
In the sci-fi world of Star Trek, in the 2370’s, the idea of a holographic, artificially intelligent physician was developed to provide short-term advanced medical assistance during emergencies in “sickbay”. This was even to the extent of literally replacing the starship’s medical officer. Yet, even in the sci-fi world, the concept was controversial. In favor of a “less human” identity, the name “Holo-Doc” was superseded by Starfleet command in favor of the Emergency Medical Holographic Program (EMH)— not very creative in my humble opinion; unlike how companies are already naming their digital workers today. By 2377, there had been four versions of the EMH, designated Mark I, II, III, and IV. The doctor, an Emergency Medical Hologram Mark I, served on the Star Trek: Voyager. Maybe you have seen him with his slightly annoying personality and questionable bedside manner (a programming issue by the creator).
Heck yeah. Resisting the EMH, however, was literally ‘futile.’ Nonetheless, some “biological” Starfleet doctors seriously objected to the use of the EMH. The familiar (and human), Doctor Beverly Crusher, herself swore that she would never use one. Nonetheless, she was forced to do so to create a distraction while the Borg (evil enemy) tried to assimilate (destroy) the USS Enterprise, allowing her, the medical staff, and importantly, their patients to escape. Yay for the EMH and rise of the healthcare provider.
Now think about this. The “doctor,” a Mark I serving on the USS Voyager, ran almost continuously for years and was forced to expand his programming extensively to keep up. Imagine all those crazy new alien diseases and life forms and new advancements in medicine? Now, the EMH itself initially objected to being the replacement for the entire medical staff. After all, the EMH program was only meant to work for about 62 days (1500 hours) after which, memory degradation would occur. Fortunately, with the help of “his” colleagues, some programming work was done to overcome this issue. AI does not sleep or take vacations, by the way.
Okay, I am a “Trekkie,” but Starfleet aside, I have also been involved in some of the coolest stuff involving AI and Healthcare. Is resistance “futile”? Well, in word, yes! Embracing AI and automation in healthcare is not a choice for a host of reasons.
The reality is we have a clinical shortage, our population is aging, we have more data, and medical discoveries are increasing at a pace that is impossible to keep up with for any clinician. Every day, in just the last 18 months, there are scores of medical journals with so many new findings. A physician specialist recently remarked that in one specialty alone, there are 800 medical journal articles with advancements in the field, which is unmanageable. I reflect back on my favorite original episodes of Star Trek when the human doctor McCoy (no AI doctors on-board the first Enterprise) was faced with healing a new species of life. Frustrated by his lack of knowledge the good doctor aka “Bones” exclaims “Dammit, Jim (Captain Kirk) I’m a doctor, not a bricklayer!”
For the first time, in 2020, we have more people over the age of 65 years old than under 5. We are running out of people to take care of an older population. Like the remote operations of deep space and sickbay, how do we support rural and urban communities when we have such a significant shortage of clinicians who are at risk of a high rate of burn out? Recall, even in the sci-fi world, the “holo-doc” complained of burnout. What is happening today to our healthcare professionals?
So, does all of this mean replacing the doctor? Well, it didn’t happen in Star Trek. Why? There is more to the doctor’s job than just diagnostic reasoning. Whether we are in that sci-fi world of 2370 or today, we know a challenge and opportunity is the right balance of technology and human elements. Even in the sci-fi world of the 2300’s, doctors had not been replaced. The job of a clinician is complicated, requiring a lot of domain knowledge, and it is a critical job that includes addressing empathy, preferences, and a range of options as well as a human touch. On the other hand, the reality is that none of us know all the things needed to care for a patient. Care is provided by a team with the expertise to provide for a patient in a particular setting and situation with a care journey that is complex enough without being interstellar (that is coming soon enough). No, we are not ready to replace clinicians, nor should we. Critical healthcare delivery is tough as it involves health – life-death consequences. It is already very challenging. So, we expand that team with AI and modern data as assets to physicians and clinical teams. We raise digital workers and integrate AI integration into workflows to ease doctors’ and operations’ burdens. Augment the clinical team? Yes.
Consider the idea of augmented intelligence and guided diagnostics. AI can use deep learning to sift through thousands of patient files and outcomes to diagnose diseases and recommend treatment plans.
Cognitive computing in healthcare will increasingly become the norm. Doctors will still provide compassionate, human-centered care, but medical AI will perform some of healthcare’s more mundane tasks; rightly so.
Consider if your doctor had treated 60,000 patients? Wouldn’t she be even better at recognizing the symptoms of a disease? Her vast experience would make her extraordinarily knowledgeable about which treatments lead to the best outcomes. Today, in Ophthalmology, AI does as good or better job at detecting diabetic retinopathy than human doctors. A similar story is true for melanoma and dermatologist, where the neural network is as or more effective. GE Healthcare’s intelligent X-ray does as good of a job at diagnosing a collapsed lung as a radiologist (before the image gets read). The algorithms perform well.
Expect to see more AI in both the provision of care and administration. Providers can use AI and chatbots to provide more efficient care, a better experience, more informed decisions by the care provider, care team, and care coordination. Conversational chat is almost human and is– here to stay. Already, there are examples where chatbot satisfaction in follow up oncology care is 90% percent and patient engagement at 80%.
Intelligent chatbots will increasingly do the jobs of issuing notifications, answering health-related questions, and keeping patients updated on their health. Just like in Starfleet, the “EMH’s will be on call day and night, ready to answer the call. Doing more straightforward tasks, customer-service chatbots will free up their human counterparts to handle more complex and time-intensive demands but also increase coverage and access.
For providers, AI can help extend the care being provided by checking on how patients are doing wherever they are in their journey: preparing for, recovering from, procedure. This already means helping predict high readmission, such as with heart failure, even predicting readmit at discharge so that the clinical team can act to decrease this event before it happens. Risks for infection like sepsis can be identified, and after leaving the hospital, help the care team manage ongoing issues. By providing more in-depth insight and leveraging data from EHR, claims and social determinants of health (where people live, shop, and eat; who they live with, etc. – not just clinical data in EHR), the care team can provide help with the toughest parts of care. The care team can change patient behavior, assess the risk, and mitigate the risk in a timely manner. Providers can also prioritize by focusing more effectively on specific populations such as a critical 200 vs. 11,000 patients for readmission. Care teams can also achieve a better-shared understanding of what is going on with a patient to reduce variation in care. Then, we also have the demands for personalized medicine, which really will require AI. When it comes to genetics, genetic testing, and primary care, the front line is somewhat lost in a sea of new technologies and advancements. AI can help do a better of job empowering what the primary care physician – from telegenetics, telemedicine to test selection and training. From a payer perspective, AI can help identify those patients not likely to close the care gap as governed by HEDIS measures. Care managers can prioritize the right outreach for preventive wellness exams.
On the whole, this is a super exciting time as we “boldly go where no one has gone before” in the future of medicine and healthcare. There are many medical, technical and ethical challenges ahead. We must aggressively address ethics and principles to maintain trust, patient safety. Trust is paramount. We need to be able to trust data used for AI (trace lineage and data from a trusted source, and whiten the “black box” for reproducibility as safety and trust mechanisms) and ensure to guard for bias.
While clinicians will always still provide compassionate, human-centered care, medical AI will perform some of the healthcare’s more mundane tasks and help clinical and operations teams manage their increasingly complex tasks while removing unneeded waste and friction – while also relieving some pressure from our healthcare workforce. Importantly, creating medical intelligence requires getting clinicians involved in healthcare’s AI transformation. We need to be smart about how we delegate things computers are better at doing. Let’s make sure not to program our holo-doc with some of those same annoying personality characteristics as the doctor of Voyager. Will we name the holo-doc more cleverly than Starfleet’s EMH? I hope. Like all new technological developments, there are unintended consequences. We will need to continue to balance technology and the human element carefully. And, let’s face it. The healthcare industry is also an opportunity rich when it comes to faster, better, cheaper ways of doing things operationally. I mean, even at home, I no longer have to get out of bed to turn off the lights or lock my front door. Thanks, Alexa.
AI including voice and NLP and new mobile tech like 5G promises to make healthcare more affordable, more accessible, and more effective for all: better outcomes, better experience, and more access. We can’t forget what it means to be human with all the tech—and what it means to care for people; holo-doc or no.