GUEST POST from Don – AI and RX +

AI and RX +

AI is coming far faster than it should.  It knows no boundaries and has no shame.  Without boundaries it will do for Medicine what Garmin did for travel.  Faster, easier, more accurate … most of the time.  110% failure at other points.  

Garmin showed us that not paying attention could put you into a lake when the program told you to cross a bridge which NEVER was there.  Or, as I experienced in a nearby city…it pointed me to drive a walking path which crossed a 4′ wide pedestrian bridge.  Another?  Yes.  It told me to drive across a back yard, cross a stream (no bridge) and drive across another back yard to get to my destination (yes … it would have saved 1.35 miles and 9 minutes if it had worked … but …).

AI will do the same for medicine.  Input 3 symptoms and the program will tell the doctor the illness, the treatment, and tonight’s baseball score.  But the in-person doctor may have noticed something which makes these symptoms more like a zebra than a horse (uncommon).  Blotchy skin when the skin should be merely reddened, for example.  AI may not be trained to notice the difference.  The patient over-did a vitamin, and isn’t really ill.  The solution is not some prescription medicine or surgery, but to reduce that over-the-counter vitamin additive.  AI will never be foolproof (in our lifetimes).  Star Trek Voyager lifetime, maybe?

No shame and no boundaries is a bad combination.  Imagine heavyweight fighters without a referee, a bell, or ropes.  Imagine no rules about where blows could land, type of gloves, or use of knives.  Not the same contest anymore, is it!

We could hope bad results would not happen.  But it is the speed of the process that determines how corrections are made.  New technology, without intense step-by-step testing, without limits, without public critique, is inappropriate.

We are already there, by the way.  “Whisper” is an AI transcription tool used by many hospitals and other medical facilities.  The problem is that it “has hallucinations” according to its users.  That is, it makes up stuff.  It has been known to make up medical procedures which were never performed and use derogatory slang about the patient which was never typed or said by the doctor entering the data.  Do you understand?  It fakes data and it is being used as you are reading this.  That data is going into someone’s medical records (and being sent to insurance companies and government oversight agencies).  The issue is serious.

Boeing kinda proves the point.  Their newest passenger planes a few years ago were ahead of their time.  And several crashed because certain sub-programs had a tendency to make strange decisions during take-off.  Pilots had to calmly re-boot the computers while trying to fly a falling aircraft.  Some were able and had enough training, but none should have had to do it.    

DIY “ORAL REHYDRATION SOLUTION” (ORS*)

NOTE: I have no formal medical training.

I was surprised to read of an easy and inexpensive way to safely rehydrate (written by a MEDICAL PROFESSIONAL).  We are heading into flu season.  Lots of people with the flu end up in the ER.  The usual reason is dehydration.  Did you know there is a Solution (ORS*) you can make at home?  You don’t need a “Sports Drink.”  Of course, if you cannot get the person to drink OR if the person is a child or aged, take them to the ER immediately. 

With an adult you might be able to delay or avoid that trip.  DO NOT EVER offer plain water.  A dehydrated person needs a solution.  It could save a life.  [Doctors recommend you have ORS packets or Pedialite, but if you don’t have those try this…]

PREPARE:
4.5 cups clean water (1 quart + 4 oz. / 36 oz.)
½ tsp salt
2 Tbsp sugar

SIP:
3.4 fl. oz. Every 5 minutes until recovered.

HOW MUCH IS ENOUGH?:
How much?  Weight in pounds x 0.0714 = possible total pints needed (it will be a lot).  A 150 pound adult may need  10 ¾ pints [1 ¼ gal] of ORS* to fully recover.  

Continue to sip your ORS until you no longer show signs of dehydration (for example, being able to pass urine, no longer feeling thirsty, and have a strong pulse).

Examine your symptoms after drinking your ORS for 1 hour, then continue checking once every 1-2 hours as you rehydrate.
It may keep you out of the E.R. Or give you some extra time to get there.  Time to get a babysitter, to get a dog watcher, to call your employer (“I won’t be in today, going to ER”).  The worst part of the ER is how long the patient must wait until they begin giving IV fluids…so this ORS* can reduce the severity while waiting.

If your patient will not drink just take him to the E.R.

[PLEASE NOTE that Don is always open to discussing the thoughts and opinions he shares here and welcomes comments as shared in the comment section. He doesn’t use other social media platforms and won’t see whatever you’d like to share with him if you post it elsewhere.
ALSO, Don is always open to offer his thoughts on various topics. If you have a specific request, you can let him know in a comment; he reads – and replies to – them all. ~ Sherry]

See more GUEST POSTS from Don!

3 thoughts on “GUEST POST from Don – AI and RX +

  1. Very interesting stuff here! I will never trust AI and, though it is getting more and more difficult and unavoidable I avoid such interactions whenever possible.
    Your ORS recipe seems to be a fairly common one across medical sites online, I’d never heard it before but I also found several alternatives at https://med.virginia.edu/ginutrition/wp-content/uploads/sites/199/2023/12/Homemade-Oral-Rehydration-Solutions-11-2023.pdf that may be more palatable for some?

    1. Thanks for your input, Owly.

      The formula came from a medical doctor who dealt with impoverished folk in places far from medical facilities. Those folk couldn’t access medical care, and many would die if they didn’t have an inexpensive way to deal with dehydration. This is a low-cost solution for such (and for wealthy people, too, strangely.)

  2. New article I read today: ”
    Doctors Say AI Is Introducing Slop Into Patient Care
    Early testing demonstrates results that could be disastrous for patients.
    By Thomas Maxwell Published December 28, 2024″
    Just one quote from the article: “Making mistakes in healthcare can kill people. Daneshjou told the Post she red-teamed ChatGPT with 80 others, including both computer scientists and physicians posing medical questions to ChatGPT, and found it offered dangerous responses twenty percent of the time. ”
    Not inane responses, but DANGEROUS responses 20% of the time. In a typical Doctor’s office which sees 30+ patients that is potentially 6 deaths per day due to bad medical choices.

Leave a Reply

Your email address will not be published. Required fields are marked *