Would you prefer a doctor fresh out of medical school or a doctor with 15 years of experience? Given the choice, most patients would choose the doctor who has logged more hours in the exam room. These experienced physicians have treated hundreds — perhaps even thousands — of people. So when a patient presents with certain symptoms, they can easily diagnose a specific disease or condition.
But what if your doctor had treated 40,000 patients? Wouldn’t they be even better at spotting the symptoms of disease? Their vast experience would make them extremely knowledgeable about which treatments lead to the best outcomes.
This theory is the driving force behind medical AI (artificial intelligence) — the same technology that powers self-driving cars and voice-activated personal assistants, like Siri.
In the next few years, you’ll likely come across medical AI that uses deep learning to sift through thousands of patient files and outcomes to diagnose diseases and recommend treatment plans.
In the last year alone, researchers have unveiled computer systems that can diagnose diabetic eye disease, skin cancer and even heart arrhythmias. In the near future, skin cancer may be diagnosed via an app that examines skin selfies. And heart arrhythmias could be detected and monitored through electrocardiogram sensors in smartphones or other wearable technology.
Hello, Doctor Robot
Chatbots will take center stage as the first non-human health care workers. Though complex algorithms will determine when these robots say things, humans will determine what they say.
Perhaps that’s why tech companies are hiring teams of playwrights, poets and novelists to write phrases for robots that don’t sound, well, robotic. With a focus on the nuances of human speech, Microsoft’s Cortana personal assistant got its personality from a team of 22 writers.
Regardless of how it’s programmed, one thing is clear. Users have an emotional response when talking to AI. As companies seek out Hollywood creatives to pen scripts for their chatbots, writers will lend life, personality and even branding touches to these soon-to-be ubiquitous assistants.
In the future, chatbots will issue notifications, answer health-related questions and keep you updated on your own personal health. They’ll be on call day and night, ready to answer any medical questions you may have. By covering these simple tasks, customer-service chatbots will free up their human counterparts to handle more complex and time-intensive demands.
The Brave New Future of Medicine
Watson first made headlines when it wiped the floor with “Jeopardy!” champions during a widely publicized matchup in 2011, but these days IBM’s supercomputer has more serious things in mind. Watson for Oncology helps physicians identify personalized, evidence-based cancer care options. The supercomputer’s cancer treatment recommendations matched those of doctors 96 percent of the time for lung cancer, 81 percent for colon cancer and 93 percent for rectal cancer.
Watson’s Clinical Trials Matching evaluates data from patient records and doctor’s notes to find patients eligible for clinical trials — reducing clinical screening time by 80 percent. And 90 percent of nurses in the field who use Watson now follow its guidance.
While Watson won’t replace doctors anytime soon, cognitive computing in healthcare will increasingly become the norm. These medical assistants have unlimited data storage, they don’t require sleep and they are free of biases that could cloud treatment recommendations. Doctors will still provide compassionate, human-centered care, but medical AI will perform some of health care’s more mundane tasks.
AI Can Spot Invisible Symptoms
While medical AI like Watson excels at recommending evidence-based treatment protocols for certain conditions, machine learning can even detect the earliest symptoms of disease.
A new machine-learning tool, created by Harvard and University of Vermont researchers, scanned Instagram profiles to identify clinically depressed users with 70 percent accuracy.
The program analyzed aspects of users’ photos like color, brightness, metadata and face detection to diagnose depression better than physician-led, in-person assessments, which had a 42 percent accuracy rate.
The study determined that depressed individuals preferred no filters or darker, grayer colors. Their photos also featured fewer people and their posts received more comments but fewer likes.
Critics of cognitive computing in medicine fear that robots will take the human touch out of an increasingly dehumanized industry. But in a country plagued by soaring health care costs, medical AI costs a fraction of what a human doctor does. There are no waiting lists. Robots never get tired, or sick or need sleep. And best of all, they work anywhere the internet does. Rather than taking humanity out of medicine, AI could make healthcare more affordable and accessible to all.
Like any new technology, artificial intelligence is not without its downsides. For starters, the revolution may not live up to the hype. Massive quantities of carefully curated medical data are needed to train medical AI. For the time being, more limited disease-targeted systems that detect conditions like skin cancer will prevail.
Since cognitive computing systems run in the cloud, privacy and security will be an ongoing issue, especially for diseases that require pictures of a patient’s face for diagnostic purposes.
And then there is accountability. Who will be responsible for a medical error if no doctor is involved? Deep learning is a mysterious process, and in many cases there would be no way to pinpoint how a machine made its error.
Brookdale Welcomes the Future
While there are many medical, technical and ethical challenges ahead, AI may make healthcare more affordable, more accessible and more effective for all.
And it may soon become a powerful new resource at a Brookdale location near you. Perhaps one day a nurse chatbot will answer your questions about senior living, or reassure you that a specific ache you have is just a minor muscle strain and nothing more.