Let's Make Personalized Healthcare Tools?
AI is the Breakthrough to end Medicine’s Dark Ages.
This weekend, I fell, I broke my leg, and I then wrote software to help me heal. I think that’s the future of modern medicine, but first, let’s start with the story…
So picture this: last weekend, I’m shredding the woods on skis with my dog. Uphill’s a party, but the descent? 90% awesome, and then ice lurking under fresh snow and a rogue bump cause me to fly off the beaten path. It wasn’t a big tree and I just slightly scraped it, but my leg got twisted back with a nauseating crunch (literally), and I sprawled in a snowbank, yelling curses to the squirrels. It hurt, but I could hobble, so I strapped my skis to my pack and limped down the mountain, one-and-a-half-legging it to my car.
Video of dog and I skiing shortly before the crash: https://www.instagram.com/kirillzubovsky/reel/DHhQymIy2f7/
So far, so good. Just me, my dog, and the quiet woods. But as I walk down, I’ve got time to kill. It’s a 30 second ski, or a 20 minute limp. I called my wife to break the news and started grilling Grok about my symptoms. I mean, of course I did, what else was there to do? I explained the details—fall, twist, pain—and within minutes, it tossed out some possibilities: ligament tears, fractures, the works. The forecast was between nothing and months of recovery. By the time I hit the doctor’s office, I was basically ready to co-diagnose. She did an assessment and sent me for an MRI. That’s where the adventure really began – the medical imaging process that followed was a time capsule to prehistoric times.
It cost $350 and an hour for an MRI where I was expecting (hoping?) sci-fi 3D clarity. Nope, suckers! I got a stack of grainy, black-and-white 2D pics—worse than a thrift-store Polaroid—and a CD. A CD in 2025! With Windows-only software and pictures in a binary format? What?
I’m a Mac guy, so that disc was about as useful as a paperweight. I had to go look at my storage for the external mac CD player from 2005, just to get the files onto my laptop (yes, with a broken leg). Lucky for me, I didn’t throw it away. Fed up though, I teamed up with Grok and Cursor to hack together my own MRI viewer. Took an hour to mostly wrestle with bugs and to write the README file, and boom—I was flipping through my knee pics like a pro. No medical degree, just a dude done with the nonsense.
Those images, though? Wild to stare at your own insides and know the answers are there, but they’re locked behind blurry pixels. Can you believe doctors make life-or-death calls off this junk?
The radiologist’s report finally arrived by the evening, a jumble of medical gibberish. I could tell my knee had some problems, but now I was also stuck with a problem of decoding the results. As you may have guessed by now, I chucked it at Grok and begged for plain English. It hedged—“I’m not a doctor!”—but I didn’t care; it was my MRI results. We hashed it out, and I knew my injury before my doc even rang to explain them. Here’s the zinger: she was so focused on ligaments, she missed a fractured bone. Not a biggie, I don’t hold it against her, but still—a break’s a break! Grok flagged it like a hawk.
This is why I trust AI over human doctors. Doctors are biased and tire; AI uses unbreakable logic, processes data instantly, and even builds tools to check itself—something no doctor can do.
It is simply logical to let computers do logic.
While booking an ortho consult, I leaned harder into Grok. It asked it to look into non-invasive treatments and found out about red laser therapy, and plasma injections—stuff I’d never heard of. It even whipped up a recovery menu, filtered out for my allergies, and sorted by meal time: bone-boosting breakfast, collagen-packed dinner..etc. Grok was like a med-school friend on demand.
This was my prompt to Grok:
I am allergic to dairy, eggs, and gluten. Can you compile a list of foods that could speed up bone, collagen, and ligament healing. Order them by most likely to be breakfast, lunch, dinner, and snack?
So here’s the deal, it is 2025, and it’s downright bonkers that healthcare still feels like it’s stumbling around in medieval times. We’ve got AI that can spot diseases, predict outcomes, and even nudge surgeons’ hands, but the tools we’re handed? It’s like the whole of healthcare is trudging through a marathon in steel-toed boots while AI is sprinting past in Nikes. We should use AI to save sanity and time, for both patients and their doctors.
My DicomViewer was a spur of the moment project. I just wanted to make it because I was handed a CD full of stones and I needed to know what was there. However, now that I made it, I can see a million possibilities. We can merge all the images into a 3D picture and then map it to radiology reports to make sure it’s clear what’s what. We can get AI to process all of that and to validate, or to question the analysis. We could make all of this patient-friendly, free, and instant.
As an engineer, I see AI as the Age of Enlightenment. I want to be able to envision and make the software, hardware, and tools that will get us through the next 1000 years. Before we could only have dreamed of making the tools for our lifetime, but now it’s truly possible to create an impact on a previously unthinkable scale!
If you are interested, I have open sourced the DICOM viewer application here:
https://github.com/kirillzubovsky/dicom-viewer
Anyone can use it or contribute to it. If I could do this little app while laying on a couch with a broken leg, imagine the possibilities if a small group of curious minds puts their energy in the right direction? It’s limitless!
Have a freaking awesome day!
Kirill Zubovsky
p.s. If you want to help my leg heal, please share Novice.Media with a colleague or a friend. If I could get to 5,000 subscribers, my ACL will magically regrow itself.