This idea goes all the way back to ancient Greece, where the Methodic school of medicine argued that illnesses could be treated by selecting treatments automatically based on a checklist of symptoms:

The Methodic school takes it to be that once a doctor has recognized the disease a patient has for what it is, the treatment that should follow is inherently obvious. It is not a matter of inference or observation, but of an immediate knowledge.

But the mainstream consensus has always been that each disease can manifest differently in each individual, and physicians need to rely on intuition derived from experience with previous diseases to diagnose each new case. And removing the physician from the loop—even in cases where final prescription would be the same—reduces the physician’s ability to make observations of the patient and the disease that could help inform them if there are future developments.

The form to fill out would receive private health information about specific individuals stored as data on some online publicly visible centralized server somewhere, including in large part the impaired and elderly.

This should never happen. This is a horrible idea. It’s not a question of if data leaks but when.

I recently self referred to physiotherapy for an issue with my elbow.

It took me 30 minutes to answer the AI chat bots questions so it could “diagnose” me. It kept going around in circles, asked me to point on a map of the body what hurt, except the map didn’t let you select elbow just upper or lower arm, at no point did it actually ask me what’s wrong because an open question and free response would be too complicated to unpick. It keep repeating questions I had already answered but worded differently and eventually referred me for lower arm pain.

When it comes to medical advice, doctors are already bad enough to talk to, replacing them with automation is just going to lead to frustration for patients and an easy out for healthcare services when it misses something or gets a diagnosis wrong.

In the future the poor will only have access to garbage AI which will end up killing many people through unreported mistakes, while the rich will still be using human doctors. The same way that the rich send their kids to private schools that teach the traditional way, while the poor are left with the mess we have now.

Also, the idea that “no one thought of this” is just ludicrous.

At the broadest level, what you are describing already exists. It’s called a “pharmacy” here in the US, and generally includes both a selection of medicines with an acceptinly low risk of harm and diagnostic equipment that can be accurately used without formal training.

For a more electronic example, there have been numerous attempts to augment what used to be a purely manual profession over the years. Several of these are adapted and incredibly useful – any doctor I see can view a shareable electronic record of my treatment, prescriptions are electronically transmitted to a pharmacy, and there are even custom interfaces for electronically checking standards of treatment or insurance coverage.

What you won’t find in general availability are “virtual doctors” that combine an inquiry as to your problem and an application of potential solutions, because humans are absurdly complicated ongoing chemical reactions whose communication is absurdly unreliable.

Or, to put it another way: a lot of people are trying, but it’s a hell of a lot harder than you imagine.

The idea of using computers to aid in medical diagnosis is older than you are, probably older than you and me put together. The field is called Medical Decision Support (MDS). The Department of Defense invested heavily in MDS, and last I knew, MDS was a key component in Armed Forces Health Longitudinal Technology Application.

Medical knowledge is collated by accomplished medical specialists, often surgeons and/or professors. The knowledge is weighted and encoded into modules using a specialized computer language. An engine runs a given module based on initial questions, then drops or engages modules based on how the questions/answers progress. Various tests are performed along the way, e.g. Bakody Sign for a cervical radiculopathy. At the end, the software spits out something like:

• 5/5 indications for C4/C5 nerve impingement
• 1/8 indications for COPD
• 8/10 indications for douchebaggery
• suggest the [X, Y, Z] for further testing

The same company (PKC) that first developed MDS was founded by the guy who also developed the Problem-Oriented Medical Record, an early form of EMR and an efficient means to quantify medical diagnoses. I worked for him at PKC and wrote the IIS version of the engine. You can see a deeply neutered version of the software at Sharecare (I think they now call it AskMD).

We were talking about this twenty years ago.

My symptoms: I had an itch that wouldn’t go away, was tired all the time (partially from spending all night scratching until I bled and then scratching for a few hours past that point instead of sleeping, and occasionally had breathing problems. After two years of seeing doctors and getting all kinds of tests, a lymph node swole up, I had a weeklong stay in the hospital with even more doctors, and it was finally diagnosed as stage 4 cancer. A herd full of doctors couldn’t figure me out, a form would have been worse.

This sort of thing does exist, but the reason you don’t see it so much, is that its generally too imprecise, and too prone to error. For example, let’s say you’re trying to assess something related to abdominal pain. Is the pain on the left, centre, or right, or multiple? Is it closer to the front, back, or both? How does the pain feel? Etc. Etc. Etc. Thats a lot of questions, all of which can significantly affect the likelihood of different diagnoses, and which may be somewhat subjective. At the same time, people can’t be trusted to fill out the survey accurately (intentionally or not), getting worse the longer and more complicated it is. On the other hand, going through a doctor, you can in theory, assess symptoms more naturally and flexibly, while confirming their accuracy so insurance/government isn’t spending on (supposedly) unnecessary procedures.