steven lewis: artificial intelligence set to revolutionize health care
4 minute read
we apologize, but this video has failed to load.
try refreshing your browser, or
tap here to see other videos from our team.
tap here to see other videos from our team.
artificial intelligence (ai) is already as good as or better than humans at many health-care tasks.
as with all science, the strength of evidence varies, but the examples below are based on controlled studies (most of them randomized) published in peer-reviewed journals.
• self-directed, web-based cognitive behavioural therapy for depression is more effective than standard treatment. machine-read electroencephalograms successfully predict which patients with major depression will respond well to certain drugs.
• chatgpt does better than physicians at answering questions patients posted to a web-based platform. a panel of health-care professionals preferred the chatgpt responses in 79 per cent of cases. its responses were 3.6 times more likely to get a high rating on quality of information, and nearly 10 times more likely to be judged empathetic.
• machine-read chest ct scans detect more actionable lung nodules than radiologists. in one large study (5,200 patients in each arm of a randomized clinical trial), the software detected eight malignant nodules, while the radiologists detected none.
• low-income obese people are seven times more likely to lose at least five per cent of their body weight in six months if they use an app to guide behaviour.
advertisement
• nursing home residents with dementia and chronic pain sleep better overnight and less during the day after interacting with a social robot for 30 minutes a day for six weeks.
• robot-assisted surgery is as good as or slightly better than conventional surgery for an increasing number of procedures.
some of this should be no surprise. health care is built on a staggering amount of constantly evolving scientific information. even the best brain is no match for a tireless computer with a perfect memory and endless capacity to incorporate new data. (think ibm’s watson destroying jeopardy! champions.)
line judges in tennis and baseball umpires are amazingly good, but their automated replacements are superior precisely where it matters, on the close calls. so it is in disciplines based on pattern recognition, like radiology and pathology, and procedures with fine tolerances, like surgery.
but what about the human element — the trust in a physician, the observant compassion of a nurse, the encouragement of a physiotherapist — that draws many into health care in the first place? that connection gets shortchanged in today’s frenetic environment. maybe it’s due for a comeback as ai frees up time.
advertisement
or maybe not. our relationship with health-care providers is not like our relationship with family and friends. it is transactional; they provide a service to improve our health. congenial visits with familiar providers improve the patient experience. but health care is existential business, not golf with our pals.
attentive, unhurried consultations are great. raise your hand if you’re not conscious of time pressures during your visits. not every provider is a gifted communicator. who among us hasn’t had to tell our stories five times? when worried about a symptom, we want diagnosis and advice now, not next month.
an ai-generated on-call avatar that knows everything about my medical history, remembers every conversation, has unlimited time and patience, communicates on my level, and knows the science cold has a lot going for it.
it can detect subtle changes in how i look and how i talk. it can tell if i understand what i’m being told. it can probably tell me jokes tailored to my sense of humour. and ai is still on training wheels.
if smart cameras and sensors can expertly monitor hospital patients and generate corrective protocols, what becomes of medical and nursing rounds? if web-based ai can diagnose most physical and mental conditions and recommend a treatment plan, what’s the future of primary health-care clinics?
advertisement
if avatars, interactive software and robots really are better at getting us to change behaviours than humans, what will tomorrow’s experts learn and what will they be?
major disruptions lurk around every corner. automation usually replaces less-educated occupations — the assembly line workers and bank tellers. health-care ai puts professions at the top squarely in its sights — the specialties focused on a narrow spectrum of conditions, patterns and therapies. a
i might endanger their very existence, while generalists may be in for a renaissance.
maybe, who knows, what if: imagine trying to predict workforce and keep health science education curricula up to date. it’s going to look a lot like white-water rafting.
steven lewis spent 45 years as a health policy analyst and health researcher in saskatchewan and is currently adjunct professor of health policy at simon fraser university. he can be reached at slewistoon1@gmail.com.
sign up for the saskatoon starphoenix afternoon headlines newsletter. click here to subscribe.
our websites are your destination for up-to-the-minute saskatchewan news, so make sure to bookmark thestarphoenix.com and leaderpost.com. for regina leader-post newsletters click here; for saskatoon starphoenix newsletters click here
share story
