Anyone can publish on Medium per our Policies, but we don’t fact-check every story. For more info about the coronavirus, see cdc.gov.

Narrative Ai in medicine [aka the beauty of being human]

Bethany Doran
Oct 17, 2018 · 8 min read

In the corner of my mind, I heard a familiar sonata drifting into the glittering expanse of the airport terminal as I departed my flight. I hurried towards the exit, noting during my passing that the piano played autonomously. It was logical that a person would not be sitting at the piano in a location designed merely for transit to another place. Still, there was something disconcerting about the image to me. It appeared as if a ghost played the music — each key clicking down in perfect rhythm while the bench remained empty, an unnecessary remnant of a different time when a human’s presence had been necessary to direct its melodies.

I passed different signs with hastily drawn arrows denoting a mess of data all pointing towards a cryptic black box.

Neat arrows emerged from the box pointing to categories promising ‘efficiency’, ‘power’, and ‘future’.

I thought of the black box, colloquially known as artificial intelligence (Ai), analytics, or machine learning and its potential for the transformation of not only the medical field but society itself. As I walked by the literal writing on the wall, I heard the words of an Ai expert ring through my head: “none of the analytics being created are pure in the way one might envision mathematics to be. All have the inherent biases of those creating them.”

I arrived to work and entered the heavy stone archway of the hospital, erected in 1930 and bearing little resemblance to the modernity of the airport I left. Putting on my white coat, I joined the other providers, frantically rushing towards their clinical responsibilities.

A intern’s pager alarmed in the elevator and the intern silenced it with annoyance, squinting at its small screen to read the letters of text on the small screen.

My friend Ray, a physician and bioethicist had pointed out the antiquity of the medical system recently commenting, “if you ever want to hammer across the point of some of the obsolete things we are attached to in medicine, just tell a non-provider that we still carry pagers. They’re always surprised.”

I defensively held up my pager. “I love this thing. They’re reliable.”

He chuckled at my response. “See, even you. Only drug dealers and physicians still can’t let these things go. Why use a pager when we have a sophisticated machine in our pockets,” he said, holding up his phone.

I thought of his comment and wondered why we were training each successive generation of providers to be attached to the primitive technology. I held a sentimental attachment to my own pager — it reminded me of the long days and nights in the hospital that I had trained in during residency and was tightly linked to the ways I had been indoctrinated into the medical field.

I joined rounds where one of the newly minted interns removed a printed list containing the names of his patients from his white coat pocket. He adjusted his stethoscope nervously around his neck, still getting used to its weight. He cleared his throat before presenting and as his hands trembled, I realized he had been reading textbooks only weeks before about patients, but now was responsible for the lives of others.

“In summary, this is a seventy two year old female. She has a history of hypertension, hyperlipidemia, and gallstones status post laparascopic surgery in 1992. She presented to the ER after she was dizzy. She was found to have atrial fibrillation on her EKG and was treated with IV metoprolol. My plan is to continue the metoprolol, get an echocardiogram, and anticoagulate her,” he said.

My attending stopped him.

“That was good, but this time, I want you to put the paper away. You’ve met the patient. Tell me their story from memory and after what you think happened and what you want to do.”

The intern shot him a worried look.

“It’s okay, you don’t have to remember every single detail about her,” my attending encouraged.

The intern reluctantly put his notes away on the patient and took a deep breath before restarting. This time he looked up at the ceiling, and I recognized that he was visualizing her face.

“She’s a 72 year old woman. Generally pretty healthy except for hypertension and vascular disease. She had a stressful event after she got news her daughter was involved in a car accident and a dizzy spell later in the day. She was found to be in atrial fibrillation in the ED, which is a new diagnosis for her. It seems like her potassium was low because she wasn’t eating which may have also precipitated the event, and I know people who are older are at higher risk because their heart structure changes. My plan would be to get an echocardiogram to assess her heart function, make sure her electrolytes are repleted and anticoagulated her because her CHADS-VASC that we use to calculate stroke risk is elevated.

The attending smiled approvingly at him and the intern smiled back, relaxed.

We went into the patient’s room and my attending pulled up a chair beside the patient, discussing the plan with her. We listened silently as she described the strain of the day prior to the hospital and worries about her daughter before reassuring her and discussing the options for management moving forward.

The interns examined her first, methodically pressing their stethoscopes over each of the areas of her heart as we all had first learned to do in medical school. I had learned the locations to place the bell of my stethoscope by thinking of the acronym ‘All Physicians Take Money’ in my head. I watched as the interns followed the same clockwise mechanistic rotation around the sternum that I had learned, accompanying the memorized letters.

After they were finished, my attending tapped the bell of his stethoscope and gently pressed it over her chest wall as he placed a hand on her shoulder and asked her to breathe softly. I noticed that his exam was different, as he imagined the location of her heart and valves in his mind, sometimes making slight adjustments to better hear the flow of her blood to her main arteries or to assess for regurgitation or stenosis.

There were new stethoscopes coming to the market that allowed recording of heart sounds and flow, and that could store data to analyze the sounds and help determine if a patient had valvular pathology without the need for trained ears. There were new ways to integrate risk scores within the electronic health record on our phones that could use data from millions of patients in real time with as many inputs as we could think of and provide precise estimates of stroke, different than our current risk scores, limited by only a handful of inputs.

Some of the risk scores would rely on seeming black boxes, and at times give estimates of risk that we could not explain ourselves except that the computer had generated them for us and they were more accurate than ones we could create ourselves. The new technologies would not only change the way we understood, analyzed, and communicated risk to patients and diagnosed disease but would change the way we practiced medicine itself.

Still, In the quietness of the patient’s room as we felt the irregular throb of the patient’s pulse under our fingertips and sat with her to discuss the things that had caused her illness, there was something that could never be replaced by advanced risk scores or technology. It occurred after she became ‘our’ patient, and after we could picture her face in our heads when we discussed her during rounds. The laundry list of her problems written on our signout instead became a narrative of her life and disease.

We left her room and briefly went over the plan, and I helped the interns finish their work before leaving the hospital. Their pagers went off, and they stopped their work to look at the pager screen, using their cell phones to call back the numbers. I watched them with amusement, thinking of Ray’s comment.

I left the hospital and walked towards the exit, seeing the red light of another dying day glimmering through the glass doors of the hospital, as the sun sunk just beyond the horizon. There was an old man who sat in the lobby of the hospital at the piano, and as I walked towards him I could hear a different kind of music than the sonata at the airport, drifting through the archways.

I paused for a few seconds, among a small crowd of people who had gathered to watch him. His eyes were closed and he approached the piano with reverence, lost in his own world. The music was beautiful, and moving, and conveyed a part of himself. It was a part of himself bound by the shapes of the notes on the page and cadence, but framed by the experience that was his life. It made the music, played millions of times before him by countless individuals a piece that was uniquely his own. His fingers, flying across the piano keys like birds transmitted happiness and sorrow, suffering and triumph and we all listened for a moment, forgetting the places we were so intent on getting to.

He represented something that could never be captured by a computer, that great collective desire and hope and love that lives within all of us. The principles that must, above all costs, be the core that forms the basis of the next phase of what medicine is moving towards. Each of the data points stored in a computer and analyzed by an algorithm represents a patient that I have met and devoted some part of my time and life to, or that another provider within the health system has met and cared for. Each data point is a unique individual with experiences and inputs too numerous and nuanced to be included in a risk score. Each is a life that can never be captured in a perfect algorithm or in the ledgers of an account balance.

We will always be the most sophisticated computers in the room, able to pause at the sight of suffering in a way a computer cannot, experience joy with our patients, and understand the nuances of interaction that are difficult to input into algorithms. Novel approaches to risk modeling and analytics should help to guide the evidence we use to make decisions and will lead to unprecedented advancements in safety, implementation, and growth of capital within health systems. However, we as providers must play an active role in the creation of principles shaping the novel analytic techniques that will guide the next phase of medicine to ensure they are biased by humanistic rather than merely financial principles.

We have the opportunity to reimagine what appropriate definitions of health, and even human existence are in the upcoming years. Capturing the beauty of human existence must be the reason the piano exists, rather than a piano bench merely a cold relic of the humanity that was left behind.

Data Driven Investor

from confusion to clarity not insanity

Bethany Doran

Written by

Humanist, innovator, and cardiologist. Plus a bunch of other things geared towards changing the world. Founder at lumas.tech

Data Driven Investor

from confusion to clarity not insanity

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade