Ran an experiment with my blood panel: converted it to CSV and sent it to 2 doctors and 3 LLMs.

Not only did the LLMs match the doctor conclusions, they had more details and were easier to digest than the doctors.

We are entering a new era of accessible diagnostics.

But how do we query LLMs with sensitive data like health records without compromising privacy?

Anyone working on solutions? Local inference, encrypted computation, zero-knowledge pipelines anything else??

medicine

AI Case Test #2

Llama's responses are great. As an experienced clinician, the LLM added nothing new but I expect novice clinicians would gain a great deal by passing their patient cases through the LLM.

I am testing LLM/AI models on different medical cases I encounter.  In this case we use Llama 3.3 and further refine our System Prompt.

Venice.ai 's LLM responses are vastly improved when using the System Prompt feature.  I also opine that LLMs will provide better responses to better interviewers.

Link to the Substack post: https://open.substack.com/pub/largelanguagemedicine/p/large-language-medicine-2?r=2tr7dt&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

I use Venice.ai when passing on real cases. Being paranoid (and concerned about HIPPA) I remove all patient identifiers. This limits the LLMs memory for a specific patient tho

I've written about my experience giving AI hard cases, which reminds me I need to get my latest case published
https://warpcast.com/0x-omen.eth/0x7e0d8240

Founder of @rotki 🐦, the portfolio tracker that protects your privacy.

#Berlin. University of Tokyo graduate. Marathon runner. #ethereum developer. #birding