Category Archives: medicine

The Future of AI in Medicine

Posted on by
8

What do I know about AI?  Next to nothing.

So why would my colleagues at the American College of Physicians invite me to speak in a panel discussion on AI in primary care?  It was a mystery to me and yet I agreed, because hey, I like feeling wanted. And I trust my ACP friends.  They know me and they knew what the panel needed.

The meeting was this week.  Daughter got sick, Husband was out of town, and I chose to stay home.  I thought, ‘Nobody will miss me, I don’t know anything anyway.’  But through the wonders of technology, I was able to participate via Zoom, yay!  And I did end up contributing, as the one clinical physician panelist.  It was fun and connecting, despite my being remote.  I recognized two former colleagues during Q&A just by their voices, and another texted me from the audience after she saw me on screen.  This technology Luddite may yet be converted.

I share here a summary of my learnings and perspective on AI in medicine at this moment—I have many more thoughts about it than I realized!  I wonder how long before it all changes?  I bet not as long as I think.

Current State

Dr. Steven Lin is now my favorite clinician writer on the possibilities and pitfalls of medical AI, and in primary care, specifically.  For two excellent and concise summaries, read an internal Stanford interview from 2019 and his original article from the Journal of the American Board of Family Medicine in 2022.  Some visual highlights from the latter:

Mission: To Enhance, Not Replace, Human Care

Intelligence is complex, and none more so than the human.  We are amazing!  For simplicity’s sake let’s just divide human intelligence into cognitive and limbic—thinking and feeling.  Intuitively we probably all understand that ‘artificial intelligence’ refers to the former, activities that can be replicated using concrete and objective data and logic.  Algorithm implementation and machine learning already enhance diagnostic accuracy, clinical decision making, and risk assessment and management.  AI even delivers health coaching now, with encouraging results.  Triage, access, billing, and follow up processes will all likely improve in the coming years with AI.  I welcome the streamlining and appropriate simplification of our enragingly labyrinthine healthcare systems.

Still, patients will always need personal encounters with clinicians who provide services and care.  But when were you last able to reach your primary care provider directly and quickly?  How long do you have to wait for an appointment when you’re sick?  If you answered, ‘recently’ and ‘not long,’ you are in the vast minority.  How well does your primary care doctor know you, if you have one at all?  So many patients have given up on establishing this bedrock healthcare relationship, because the system makes it next to impossible for you to talk to me when you’re sick, my schedule is booked solid for three months so I can’t see you anyway, and when I do see you, I only have 15 minutes.  It’s like speed dating with earplugs and snorkel masks.

AI will never know and interact with humans on an emotional, relational level.  What it can do is remove the transactional, clerical, algorithmizable(!) tasks from clinicians’ plates, so we can sit and talk with you, see and know you as a whole person, and care for you in the most personalized ways.  AI can learn which of you has a lot going on medically, who needs extra time and attention. But only a human can deliver that attention in a way that truly cares for and heals you, especially if your health needs are complex.  And the more AI shows us what’s needed, I suspect we will need many, many more intellectually, emotionally, and relationally competent clinicians in all fields to answer the call, which is a whole other mountain unto itself.

Data: “Garbage In, Garbage Out“

Applying populational data to individual decision making has always been both science and art.  AI can enhance this very personalized activity by making vast amounts of data easier for clinicians to access and interpret in real time patient care.  But beware big data:  How is it collected?  How do we measure and confirm its accuracy?  How is it entered, organized, aggregated, interpreted, used?  Who decides all of these things, and what interests do they have?  Who will monitor for bias, equity, and unintended adverse effects on certain populations that are inherently, if inadvertently, built into AI systems at the level of design?  This is where I really see us falling down and not even knowing, and then ignoring and dismissing those who call it out.

Governance: Honest, Transparent, and Accountable

After we answer the question of who will monitor, we must ask and answer how.  The American healthcare system is fundamentally capitalist and increasingly consumerist, an extremely high-risk combination.  Bad actors will certainly subvert the mission of individual and collective health and well-being to make more money.  But possibly more dangerous are those who honestly believe they are out to help people and who, despite their best efforts, succumb to economic and cultural pressures to put profits (or market share, or personal status, etc.) ahead of professed altruistic mission.  Short term financial and social costs of inventing, experimenting, and iterating innovation can be intolerably high, and too often those in power and regulatory positions act on primal (though well-rationalized) instincts to forgo those costs to protect and advance other interests.  At our core, we humans are emotionally driven decision makers who then justify our actions (often vociferously). Only when we accept this reality can we hope to regulate our systems honestly, transparently, and with true accountability.  I feel deeply cynical about the likelihood of us doing this at all effectively.

Agility and Evolution: Lightning vs. Glaciers, Integrated Co-Creation, and Emergence

Technology moves increasingly at lightning speed; medical knowledge doubles now about every 4 years, compared to every 50 years in 1950.  Medical culture, on the other hand, moves at a glacial pace.  I’m thinking of gender and racial equity, holistic mental health, and mind-body awareness and practice.  To avoid harms and fully benefit from the inevitably rapid evolution of AI, we must build systems that foster invention, experimentation, and iteration at local nodes (team, practice, department, hospital, health system, etc), strengthen multidimensionally integrated and transparent internodal communication structures, and maintain monitoring processes that can detect common, coincident AI benefits as they emerge, so they can be efficiently, effectively, and appropriately scaled.  At the same time, solutions must be flexible enough to adapt and iterate on a global level, as shared learning marches onward.  This open-source and resource-intensive style of collaborative innovation feels antithetical to the competitive, short term profit-driven American healthcare business culture.  I see this challenge as especially inextricable from the accountability problem of AI governance above. Yikes.

Medical Education:  Leading and Learning by Example

Sometimes I wonder if technology is making us all dumber–like maybe safety features on modern cars makes us fundamentally less mindful drivers?  We clinicians all have Google and UpToDate at our fingertips; accessible information 24/7.  Almost every day, I look things up in real time during a patient encounter—did you know some people grow more lipomas if they drink more alcohol?  There is just too much information now to keep it all in my head.  I feel comfortable thinking out loud with patients, talking through basic physiology and how it may relate to or explain their symptoms.  When I come across something novel or atypical, as long as the patient is stable, I feel a little exhilarated, like we’re going on a themed scavenger hunt for diagnosis and treatment.  This is exactly where AI can optimize care, by gleaning the vast seas of medical information and knowledge, filling the gaps in my head that will only get larger over time.  It gives me more bandwidth to ask better questions, listen longer to patients’ answers, and explain increasingly complex plans of care better. 

Having trained on the cusp of the information age, I wonder how/whether my thought processes differ from, and how that affects my relationship with, younger docs?  Medicine has always been an apprenticeship profession, and we have always felt social generation gaps that mirror society at large. I think that will not change. What has changed–narrowed–is the information/knowledge gap. 20 years ago, my teachers consistently knew exponentially more than I did about almost everything. I think those days are long gone. Any student today who just rotated through any given medical subspecialty will know more than I do about new diagnoses, drugs, and procedures. I am totally okay with this–there is so much to learn, I should get it wherever and from whomever I can.

Teachers and mentors, however, hold experience, intuition, context, and judgment that can only come with cumulative time spent in practice. This is what we have to offer our trainees–the example of curiosity and humility for lifelong professional and personal learning. As AI accelerates and infiltrates our systems, teachers must role model critical openness, mission- and relationship-centered integration, and an agile, honest, growth mindset. None of us will ever now know it all. Optimal patient care merges expert knowledge with presence, attunement, wisdom, and connection–machines cannot do or teach this, and we must all help one another figure out how best to incorporate machines into patient care.

Cautious Optimism

Primary care has the greatest potential and privilege to see, hear, and touch every other medical specialty, and all aspects of the healthcare system at large. Wow. How could I not love this work? What fascinating and rewarding polarities to navigate–tradition and innovation, conservation and progress, intellect and relationship, strategy and vocation. I have no complete (or even well-formed) answers, only reflections. I know where my values and goals live, though, and I root down to them. If we can do this as a profession and a society, then AI’s potential to make all our lives better could be limitless.

Breaking Point

Posted on by
8

“Dr. Cheng, I feel dizzy.”

We’ll call him Joe. I miss Joe. I met him in his early 60s, a pleasant, dapper, rotund man with a jovial demeanor and well-groomed mustache. He was always on time to his appointments and came with crisp reports on his subjective state of health. Whenever I saw Joe on my clinic schedule, I knew at least that part of my day would be good.

Joe lived a conscientious lifestyle. He paid attention to food and movement, and cultivated relationships that held him up in life. When I think of Joe, I remember feeling unworried about his habits and longevity. So it bugged me that his blood pressure was never well controlled. On maximum doses of four medications, it was consistently 140/90 (ideal is 120/80 or less), and it just would not come down. Luckily for me, his previous doctor had done all of the appropriate work up, including sleep study, kidney imaging, stress testing, etc. Eventually they decided together that this was as good as it would get, and Joe would just focus on doing the healthiest things he could every day and get on with life. I concurred.

So when I got the call months later that he was dizzy, my heart sank. We agreed he would come in that day and we’d figure out next steps. I was prepared to call the emergency department and my cardiology friends to let them know I was sending him over for a cardiac event. But on arrival he didn’t look ill or unwell, just moved a little more slowly. His mood was great, though, maybe even better than usual. I asked him why, what had changed? “I retired!”

His blood pressure was 90/60. He was still taking all four of his blood pressure medications, like the diligent patient that he was.

Back then, in a typical primary care practice seeing a patient every 15 minutes, I did not consistently ask deep questions about people’s work stress and meaning, like I do now. Turns out, Joe had a lot of mental stress at work. He managed it well, though–never lashed out at people, never let it affect his performance. Colleagues threw him a big party and he retired feeling satisfied, looking forward to his next life chapter. And it wasn’t until later that he realized how much his work had cost him in health. Over the next several days, we learned that he only needed a low dose of one medication to keep his blood pressure in the 110/70s. The dizziness went away. I think he was getting ready to go on a cruise. I left the practice soon thereafter, and I don’t know about Joe now. I hope he’s still happy and well.

I remember this story so clearly because until recently, it stood out as unique–that my patient could correlate such a dramatic improvement in a crucial vital sign and a leading risk factor for heart disease to retiring from a stressful job.

Not so unique anymore, though. This year, no fewer than four of my patients have experienced the same thing, but in the opposite direction. Having never had high blood pressure before, they all called me with new symptoms: headache, fatigue, full body tension, and just feeling off. Before calling, they thought to monitor their blood pressure, and all of them reported consistent home readings well above 150/90. And they all knew exactly when it started, all correlated with severe work stress escalation. Adjuvant factors included increased travel, longer work hours on global calls, less time for exercise, and continued business eating. It’s all a downward spiral, for sure, and I submit that the underlying cause, the change that makes four people present suddenly in these six months, is our post-pandemic work environment. While I welcome the increased flexibility and autonomy of a hybrid office model, I’m not sure the net effect is good for us long term. Work-life boundaries have perforated, if they even still exist at all. Given the volume of articles in business periodicals progressing from ‘quiet quitting’ to overt labor force decimation, it’s quite clear to me that we have not invented a kinder, more effective work culture. More and more I hear and feel an ‘us versus them’ gulf growing between workers and leaders–yet another relationship domain affected by polarization. It’s as if the short term, profit driven, make money at any cost prepandemic ethos, suppressed and reassessed (I thought?) during three years of acute obligatory disruption, became an abscess that has now ruptured and resurged–exploding like ‘pus under pressure,’ as we say. Organizations scramble to restore anteCOVID earning and productivity status, still measuring success and achievement with the same metrics as before nature showed us how vulnerable we and all of our systems are. We have learned nothing.

All of these patients are now taking at least one blood pressure medication, pulling hard on their stress management skills, and maybe reassessing whether staying in their current roles is worth the cost to their health and relationships. We connect regularly and I always ask, how much longer can you keep this up? How will you know when enough is enough? What will you do then? I don’t recommend that people quit their jobs–that is not my place. But I ask the important questions, lovingly and bluntly.

What does it cost us indeed, as a society, to be killing our workers this way? When will we recognize that sacrificing people in the short term actually wrecks collective success in the long term? When will our culture value people over profit? I hold leaders accountable for their relational output at work, and I also recognize that they–as we all–are simultaneously agents and victims of our complex adaptive systems. Culture does not change easily.

If I have any hand in it, however, more of us will move closer to challenging and changing the most toxic aspects of business culture, one organization, one team, one person, one interaction at a time.

Three Years Ago Today

Posted on by
7

Where were you on February 25, 2020? What was happening around you? What were you doing, planning, looking forward to, worried about?

What day did COVID change your life?

On New Year’s Eve, 2019, my niece declared that everybody was overreacting to the infection sweeping across Wuhan, China. It won’t be a big deal, she said. I specialize neither in infectious disease nor public health, but I knew then that what was coming would be a big. fucking. deal.

This morning, stepping out into the bright sun and crisp air at the end of another unusually mild Chicago winter, Daughter and I recalled 3 years ago. We had just spent a long weekend in Seattle, just for fun. Son had said then that he could see himself living there ‘after school,’ and lo, he ended up there for school. Little did we know then that COVID had already landed, right there, where we were.

Within weeks, whole organizations and governments mobilized and immobilized; life mutated irrevocably around us everywhere. COVID hit me in the face by cracking one of my friendships early on. The week of March 9, I begged off of a gathering, and my friend was angry. I respected and admired her, and loved her family–still do. She subsequently railed at the prospect of lockdown and social distancing, citing economic fallout, apparently dismissing my alarm at the risk to our healthcare system, and my own colleagues, if we deferred such drastic measures. It took me by surprise, floored me, and I was hurt. I wanted to talk more about it, talk through it. But these three years, every time we approach our differences here, she respectfully declines to continue. I respectfully agree. We hit a boundary in our friendship then, which I accept. I can engage with difference elsewhere; there is no shortage of opportunities. Over this time I believe I have both sharpened and softened my communication skills around disagreement and dissent, and I’m still grateful for every chance to practice, learn, and improve.

Facebook showed me what I posted three years ago today. I don’t know which leader moved me to profess my appreciation on social media, but I bet it was one of two, so I emailed them both today with the screen snip of the post. “…recalling 3 years ago, watching and waiting for the pandemic wave to hit us, standing in a state of novel awe and uncertainty.  I also remember feeling confident, though; I understood the medical and public implications and trusted my immediate professional and personal circles to think and act rationally and thoughtfully… Our organization has its challenges, both intrinsic and extrinsic.  Each day, though, I’m still convinced and proud that everybody’s doing their best for the greater good. So Happy Saturday, and thank you for you leadership. 🙂” Having led a small practice through the first 14 months of the COVID morass, I suspect any sincere expression of acknowledgement and appreciation, at anytime, even now, boosts any leader’s spirits. We are all still going through it, and we still need to hold one another up.

This is the 42nd post that appears on a “COVID” search of this blog. I process by writing. Reading back, I apparently felt more fear and anxiety at the outset than I may remember. Memory is complex! Today I recall vividly the acutely discordant and uncomfortable conversations with my friend, while minimizing the overall stress of the time and circumstances? How fascinating. We would all do well to ‘remember’ this paradox of recollection as we continue to navigate, negotiate, and collaborate hereafter.

“We write to taste life twice, in the moment and in retrospect.” Anna Quindlen includes this quote by Anais Nin in her book, Write For Your Life. Have you anything to review from three years ago–emails, letters, photos with captions, blog posts? What does any of it show you about your feelings, thoughts, actions, and relationships back then? How has your life evolved and transformed in the short and profound time since? How have your relationships moved? I’m gratified to take some time today to recall and reflect, and to have concrete evidence of myself to do it with.

Write on, my friends.