Your AI Doctor Will See You Now
Posted July 02, 2024
Chris Campbell
Once upon a time, a doctor would come to your house.
He would walk inside, smell the air, feel the temperature and humidity.
He would, without trying, notice things. Big things. Small things. Clean things. Dirty things. The environment would bombard the senses with millions of microperceptions.
And then he would examine the patient.
In those days, the environment in which the organism (you) live was not separate from the health of the organism.
Since I have an irrational hatred for the word “holistic”...
We’ll say they took an integrated approach.
Fast-forward to now…
70-30 Split
It’s called the "70-30 split."
Doctors spend 70% of their time collecting information and only 30% making decisions and communicating with patients.
It's a phenomenon so common, so customary, it’s considered a universal constant in the medical profession.
Like E=MC2… the second law of thermodynamics… gravity… awkward silences in elevators.
But the kicker: the information they're gathering has little to do with, in practical terms, helping the patient get better.
Today's doctors are drowning in a sea of digital data.
They're spending less time on the patient and more time clicking through endless electronic health records (EHRs), scrolling through lab results, and deciphering medical codes.
There’s even a phrase for it - “click fatigue” - showing how bad it’s gotten:
“The experience of physical and mental exhaustion due to repetitive mouse clicks and keyboard typing.”
I’m told the ridiculousness is unparalleled.
EHRs - introduced with the noble goal of streamlining patient care - have become a labyrinth of multiple screens, menus, buttons, forms, and flip-flaps.
SOAP… HL7… FHIR… ICD-10… CCD… CPOE… EOB… PACS… MU… HIE… ADT…
Healthcare practitioners have to learn and remember so many acronyms and buzzwords…
You’d think they’re gunning for a job at the Department of Defense (also known in some circles as the Widespread Acronym Spawning and Tracking Enterprise (WASTE).)
And the biggest mismatch:
Sure, there’s data involved, but helping humans is heavily a “right-brain” job.
It involves intuition, creativity, and other “soft skills”. Things that people who like to click buttons and tap keyboards all day generally lack.
Somehow, we’ve turned this art - the ancient art of healing people - into a formulaic, robotic discipline. Ironically, as Dr. Conor Judge has mentioned, robots can help flip things right side up.
Can AI Help… Really?
So, what’s the solution?
Dr. Conor proposes a fresh perspective: the responsible use of what’s called “multimodal AI”.
Unlike single-modal AI, which processes one type of data, multimodal AI integrates various data forms - text, images, numbers.
Dr. Conor argues this mirrors the multifaceted approach doctors use in their practice: using all of the senses.
We’ve all heard of the AI tools like ChestLink that can autonomously read scans and pick out anomalies with uncanny accuracy (sometimes beating out humans).
Those are expanding at a rate becoming impossible to track - and will soon become commonplace.
The new flavor of the week is called Eureka - a personalized AI doctor that specializes in thyroid care.
It’s the first AI doctor covered by health insurance. It can order labs, it’s 90x faster than most care in the US, and most people want to continue using it afterwards.
Multimodal AI will supercharge these AI systems - allowing them to take in data from all sources.
A Return Home
The art of human-to-human medicine can’t be lost.
Doctors develop the ability to "eyeball" patients, assessing their condition at a glance—a skill that sophisticated models can’t replicate.
Humans are much better at the integrated approach, however useful AI models become.
And, slowly, healthcare’s coming back home.
Cheap, in-home sensors will allow doctors to track data and biomarkers in ways previously impossible.
While I’m not overly keen on the “smart gadget” revolution…
Toilets that track your biomarkers… an air purifier that tracks your air quality… a robo-vacuum that tracks and alerts of potential toxins…
These all sound like good ideas - and AI can help make sense of it all.
The future of AI in medicine should fulfill the original promise of digitization: enhancing healthcare efficiency, personalization, and accessibility.
And, by the looks of it, AI can go well beyond that, too.
Remote areas will be able to gain the same insights as world-class cities, bridging gaps in specialized care.
Most important:
The AI-human partnership has the potential to allow doctors to spend more time with patients.
If AI can flip the 70-30 split and kill the click fatigue…
That’s a rosy picture of the future of healthcare.
But it’s a realistic one, too.
More on our beat - how to invest - tomorrow.
Stay tuned.