intelligence - is it any way intelligent at all (NEW CONCEPTS)

[zainalara]

The problem of undifferentiated patient populations

Pre-hospital care, primary care and care delivered in emergency departments involves the fullest possible range of clinical diagnostic acumen, as these settings provide advice and treatment on completely undifferentiated patients. That is, patients with no diagnostic pre-filter.

People who ring 999 or turn up at a GP practice or Accident and Emergency with no predesignated diagnostic label pointing their first contact care clinicians to a specific diagnosis or narrowed range of options.

Edging towards a clear diagnosis in these undifferentiated patient populations, is often therefore very challenging particularly where an ageing demographic (often with a range of co-morbidities) present to the clinician.

Where does AI fit in?

Faced with these difficulties it is unsurprising that science and the lay community look to computing advances for assistance. The dawn of precision medicine, pharmacogenomics, and especially artificial intelligence (AI) seem to offer new tools for the struggling clinician in first contact care.

Indeed, early reports seemed to suggest that we could one day dispense with the primary care clinician and instead rely on super-specialists to deliver ‘cures’ once the AI engine had made a definitive and accurate diagnosis.



AI can be defined as technologies which can perform specific tasks quicker or better than humans. A subdivision of AI is the concept of machine learning (ML). Which is the application of algorithms to recognize complex relationships or patterns from empirical data, in order to make accurate decisions.


Using ML, a computer is provided large amounts of data and a set of algorithms in order to perform a task. The data then reinforces correct answers, so that continued education of the machine occurs with no additional programming.


Current AI applications are usually to be found in defined diagnostic groups to direct therapeutic advantage, for example refining the assessment of changes in diabetic retinopathy over time or machine learning enhancing the diagnostic accuracy of potential breast carcinomas on mammography.

These are focused and specific to predefined patients with a clearer and more linear diagnostic question in place. Such as ‘has this person got a breast cancer?’ Yes or no. Has this person’s diabetic retinopathy worsened since the last image was taken? Yes or no.


These areas of medicine are clearly important, and this use of AI could have a definitive and positive impact on these patient groups. However, even within these narrow areas of clinical activity issues of rare disease prevalence creates a significant boundary for AI. Other issues of reliability, and the potential hidden nature of the decision made by the machine, so-called ‘black box decision-making’ have raised concerns more widely about the applicability of AI in medicine.

[Wile E. Coyote]

Hello Zainalara.

Wiley's view is that most decisions now are a mix of silicon and human intelligence.

Wiley's personal preference is for Monte Carlo based diagnostics, ie you use repeated random sampling to solve problems, these generate a range of possible outcomes which can then be analyzed to provide a probabilistic prediction or approximation.

Your patient would be told "we don't know", it's roughly a 14% chance of this, a 42% chance of that, etc etc. If it's this, you have an 87% chance of making a full recovery, etc..

From this your patient can then make a decision.

[Wile E. Coyote]

The Doctor's role remains to provide Occam's Razor......for the understandably anxious patient.

[Mick Harper]

Where's Boreades when you need him? Our man at the medical rockface.

[Boreades]

[Boreades]

An on-topic response.

The use of AI in medical and clinical situations is proving rather divisive.

Some of my medical colleagues are huge fanboys and think it's the best thing since sliced bread. Coincidentally, they are also at the younger end of the age spectrum.

Older colleagues, including myself, tend to be more cautious. We have tried a number of tests on the efficacy of AI results. A few lessons have been learned. In no particular order:

1) ChatGPT (etc) - they are not search engines.

2) Given any topic, they have a great ability to scrap content from other websites. The results produced are usually the ones that scored highest in rankings or popularity on the other website. But that hides (3)

3) Age and conformity. It takes time for any source item to accumulate high scores in rankings or popularity. That inevitably means the top results tend to be the oldest, or so old they are no longer valid, or the ones that most conform to the consensus. Which tends to hide newer research or work that does not conform to the consensus.

4) Intellectual property - ChatGPT (etc) are notoriously bad at ignoring any copyright on source material

5) People trust the results too much and don't go to source documents, or don't do their own research.

6) Perhaps the most serious point of all - AI results are prone to hallucinations. That can manifest in many ways. One way is AI quoting sources that simply do not exist. Some people have already been caught out by this, simply copying and pasting AI results. Which look plausible until looked at by a Subject Matter Expert (SME).

Is that enough for starters? If I think of any more, I will come back.

Edit:
I have heard stories of people gaming AI results by massively up-voting a topic in favour of one side of the discussion. But this may be urban myth. I will try and find something to substantiate that.

[Boreades]

Many legal hallucinations have already been made manifest. Do your own research.