TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Towards accurate differential diagnosis with large language models

105 pointsby gwintrobover 1 year ago

8 comments

doixover 1 year ago
This doesn&#x27;t really surprise me given my experience with doctors. If it&#x27;s a common problem then doctors do fine. As soon as you&#x27;re not the common case or have something somewhat rare, it becomes a bit of a crap shoot, at least in the UK when you deal primarily with your GP (general practitioner).<p>I have a friend with Crohn&#x27;s who was feeling low energy. I was a gym bro at the time and convinced him to take a testosterone test (because all problems re caused by low T when you&#x27;re a gym bro).<p>His doctor wouldn&#x27;t even entertain the idea, saying he&#x27;s a young man and it&#x27;s very unlikely that he&#x27;d have low T. He did the test privately and his T is significantly below normal. If you Google, there are actually many papers showing correlation between Crohn&#x27;s and low T. I bet an AI would find it.<p>Similarly, doctors missed my mum&#x27;s recent cancer diagnosis. She also had factors that would make her more susceptible to breast cancer, googling finds many papers that show causation.<p>The problem is that those things aren&#x27;t extremely common and haven&#x27;t made their way to NICE guidelines or whatever GPs use.<p>Not that I&#x27;m blaming doctors, they have 10 minute appointments and don&#x27;t have time to do anything. I&#x27;m sure AI would recommend significantly more lab tests which would put even more pressure on the NHS.
评论 #38527361 未加载
评论 #38527095 未加载
评论 #38527606 未加载
评论 #38526527 未加载
restersover 1 year ago
If it&#x27;s not already obvious, LLMs are going to be doing most of the mental work currently performed by doctors, lawyers, accountants, etc.<p>I have already nearly stopped using Google search for anything, in favor of GPT-4.<p>GPT-4 has helped me very quickly prototype things that I normally would have had to spend hours researching.<p>GPT-4 has also created custom curriculum for me to help me learn various things for which I have struggled to find good books&#x2F;tutorials online.<p>There will always be many areas in which a solid human intellect and well-honed human judgment is still useful, but much of the less critical work will yield to LLMs.<p>If we call the difference between GPT-3 and GPT4 1x, then I would expect to see a 2x-5x improvement in LLM capability within the next few years just based on how much great work has recently gone into shrinking really big models so they can run on smaller hardware.<p>LLMs are not digital human minds, they are simply very good information synthesizers. Information synthesis happens to be what most white collar professionals get paid to do with their brains.
评论 #38525741 未加载
评论 #38525732 未加载
评论 #38527757 未加载
评论 #38525634 未加载
评论 #38530135 未加载
评论 #38526081 未加载
评论 #38525606 未加载
rickysahuover 1 year ago
We are doing similar work at GenHealth.ai and getting sota results on some evals (not yet published). Our approach is very different from LLMs in that we are using a medical coding vocabulary and we are training transformers on actual patient histories. We have an API if anyone here wants to build on it. Oh and we are hiring
rrspover 1 year ago
I wonder if they made any effort to check whether the NEJM case studies that this whole study is based are in the PALM-2 training dataset.
评论 #38526106 未加载
评论 #38525779 未加载
评论 #38525834 未加载
hackerlightover 1 year ago
In poor countries there aren&#x27;t enough doctors. Billions of people don&#x27;t have access to healthcare and that won&#x27;t change until the country becomes wealthier. Public healthcare that covers everyone <i>can&#x27;t</i> happen because the reality of poverty makes it impossible. For these billions of people, the valid comparison isn&#x27;t human doctor vs AI, it&#x27;s no treatment vs AI. I am extremely optimistic about the positive role of AI in medicine for this reason.
评论 #38527106 未加载
gcanyonover 1 year ago
Sure, but how will they train in an appropriate level of snark, abuse, and sexism to convey authority?<p>(I miss House and hate how it ended)
westurnerover 1 year ago
Differential diagnosis &gt; Machine differential diagnosis: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Differential_diagnosis" rel="nofollow noreferrer">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Differential_diagnosis</a><p>CDSS: Clinical Decision Support System: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Clinical_decision_support_system" rel="nofollow noreferrer">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Clinical_decision_support_syst...</a><p>Treatment decision support: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Treatment_decision_support" rel="nofollow noreferrer">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Treatment_decision_support</a> :<p>&gt; <i>Treatment decision support consists of the tools and processes used to enhance medical patients’ healthcare decision-making. The term differs from clinical decision support, in that clinical decision support tools are aimed at medical professionals, while treatment decision support tools empower the people who will receive the treatments</i><p>AI in healthcare: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Artificial_intelligence_in_healthcare" rel="nofollow noreferrer">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Artificial_intelligence_in_hea...</a>
评论 #38526551 未加载
guenderover 1 year ago
Yikes, do I read this correctly and the LLM alone outperforms clinician + LLM?
评论 #38525613 未加载
评论 #38525860 未加载
评论 #38525944 未加载
评论 #38525652 未加载
评论 #38525902 未加载
评论 #38537651 未加载
评论 #38525501 未加载