top of page
Writer's pictureBurt Rosen

Dr. ChatGPT vs. Dr. Google (spoiler: Dr. ChatGPT is going to win by a lot)

Updated: Oct 3


I might be a little late to this party (I have high expectations of myself) but I am currently obsessed with ChatGPT. Haven't tried it yet? Definitely do. Go ask it any questions and ask it to respond to you in any voice you want.


Are you asking for an analysis of data or medical info? Ask it to translate it into plain English. Do you want it funny? Ask it to write it as comedy. It's pretty amazing what you get back.


I am speaking on a panel soon about AI in healthcare so I have been testing ChatGPT a lot and thinking about ways that it can be used to help patients.


For fun, I was wondering how it would do interpreting my scans in case my scan appointments and doctor visits are far apart. I cut and pasted my last CT Scan report and uploaded it and asked for an English explanation. It was dead on and easy to understand.


Here's an excerpt that ChatGPT wrote in English as I had requested (I am transparent so I hope you are ok with me sharing this). (Comparison of versions was a great suggestion by my friend, the legendary ePatient Dave (deBronkart)):


You can see how easy it is to understand.


So how else can I see AI being used in Healthcare? A ton of ways but some quick thoughts (I have more every day):

  • Pulling together data sets from multiple sources and making sense of them

    • Adding an integrative oncology treatment data set and a medical treatment data sets nd looking for causation, correlation a,nd trends

    • Building integrated treatment plans

  • Predicting outcomes and results based on past data and reactions to treatments

  • Filling in gaps in med school education - students can't learn everything so AI can help fill in gaps

  • Overall trends - I keep track of my symptoms every day. I uploaded my data into ChatGPT and it found relationships between my symptoms

  • Reading of test results and translating into plain englishE- also trending and charting


It's really cool but far from perfect, so take it with a grain of salt.

  • The output is based on the data it's reading so garbage in, garbage out as we say.

  • Data sets don't allow for a lot of diversity of people since there are groups whose data isn't as good as others

  • The power will ultimately be in the ability to read every detail and predict outcomes based on modeling of YOUR personalized data


I am beyond excited about the possibilities. Precision and individualized oncology is the holy grail and using AI to understand all about Burt will be much more effective than treating me based on averages and what works for a % of others.


Stay tuned. These are just my early thoughts!


50 views0 comments

Comments


bottom of page