| HN Mirror

I'm about to graduate from medical school in a few months. Once I get my medical degree, I'm going for an undergraduate degree in Computer Science. I've always wanted to do computer science since high school but I equally wanted to become a doctor. At one point I decided being a doctor was more important. I valued autonomy highly and I figured nothing is better than the freedom to operate my own small private practice in the future. I didn't want to end up being employed as a programmer and I expected that being a freelancer is no where near as stable in terms of job prospects as being a doctor, especially on the long run. Add to that the fact that I was fascinated by how the human body works for the same reasons I was fascinated with computers. I think any hacker-minded person would. I wanted to know the ins and outs of the human body. I wanted to know how it breaks, and how to fix it. Also, to be honest, I was very tempted with the extra income that comes with medical practice, but autonomy was ultimately the primary motive.

6 Years later, here I am, graduating with a decent GPA but having 0% interest in pursuing clinical practice, although i have performed very well clinically and in terms of my medical knowledge. I simply realized that the practice of medicine (specifically diagnostics and treatment decisions) is nothing but a classification problem. We are literally trained to memorize 'algorithms' (flow charts) for diagnosing and managing hundreds of different illnesses. That's about it really. It boils down to asking a standard set of questions (the patient's history), examining the patient, and trying to guess the diagnosis. Often, lab tests are needed (you order them based on the flow chart you memorized for the presenting symptoms). Once they're available, the diagnosis is usually clear, or further testing and imaging is required, and the cycle continues. Even for the complicated medical issues, this whole process can be represented by a simple flowchart that easily fits on an A4 sheet). The majority of modern day work done by doctors can (and will) be automated in the near future. The biggest hurdle was never the technology; it's the laws and regulations. I've seen papers published in the 80s where AI bested human doctors in diagnosing many diseases, and recently I've seen more impressive results in radiology and pathology diagnostics, where the diagnosis purely depends on vision, which is a very complicated problem when you try to solve it with computers. The results are very promising and some papers have shown results where human pathologists and radiologists were outperformed by computer vision.

Humans are obviously still needed to deal with patients. It takes some clinical skill to know how to extract information from patients, how to examine them, and how to look for clues, but that does not strictly require a doctor to be done properly. In fact, most of our medical training, even in residency, is concerned with learning more and more 'algorithms', guidelines, and staying up to date with the latest medical evidence.

Computers won't replace doctors in surgical specialties, but they certainly will replace primary care physicians, as well as doctors in other fields like internal medicine (including its subspecialties like cardiology, Pulmonology, etc), emergency medicine, oncology, and others. These fields purely depend on memorizing and recalling flow charts of diagnosis and management. Humans are fallible when it comes to memory and recall, computers are much much less fallible in these cases, and virtually infallible in some cases (100% diagnostic sensitivity and specificity was reported in some studies, which is impossible to achieve by humans)

By the time I realized what I mentioned above, I was already close to finishing med school. I decided to graduate first then see what I wanted to do next. I no longer saw any inherent joy or value in being a doctor. There is no room for trying to be creative, smart, or efficient. You just have to follow the official guidelines and policies, and hope you don't get sued when you, inevitably, make a mistake. I see no meaning in a career like that.

I love coding and I love learning CS in my free time so my decision for what to do next was easy. There are some interesting studies utilizing AI for diagnosing cancer metastasis on CT scans using computer vision, and that's an example of a topic I might be interested in. I don't necessarily want to do medically related research all the time though. All I want to do is to be a software engineer working on very interesting projects. Even the most mundane coding projects are easily 10x more mentally stimulating than clinical work in my opinion.

In retrospect, I have no regrets. I know a lot about how my body works and I leanred a lot from dealing with hundreds of patients over the years. I also learned a lot about who I am and what I really want, and to me, this knowledge is invaluable. I also realized how much I appreciated computer science.

I guess that's enough rambling. Apologies for the late and long reply. I hope you found it useful, and I wish you the best of luck in your future career ;)