NotBiebs and 69 others on Nostr: I made a privacy focused app for analyzing blood work. It's similar to , but simpler. ...
I made a privacy focused app for analyzing blood work. It's similar to
https://getbased.health, but simpler. Currently, you have to run the app locally and also have Ollama running on your machine. So there's more steps to get the app up and running, but the privacy guarantees are better and you can run the app completely offline once you have everything set up. The app gives you insights about specific lab panels as well as trends in your overall health and specific health markers. You can also chat about a specific lab panel or trends of a specific health marker. I've been using gemma4:26b-mlx with the app which works great on my Mac Mini M4 Pro with 64 GB of memory. I was originally using qwen3.6:27b which gave me good insights, but was really slow. gemma4:26b-mlx is a lot faster and generates responses similar to what I was getting from qwen. The repo includes a /compare-live-evals skill that can be used to help you choose what model works best on your machine. I currently only have live evals for panel chats though. I'm planning to add live evals for trend chats soon.
https://github.com/SamSamskies/locallabPublished at
2026-07-17 19:16:51 UTCEvent JSON
{
"id": "1945f3f6c2fdb3a5768f90c5c88b4ef6fdf8815cba158b90d94047159cf8b6ca",
"pubkey": "604e96e099936a104883958b040b47672e0f048c98ac793f37ffe4c720279eb2",
"created_at": 1784315811,
"kind": 1,
"tags": [],
"content": "I made a privacy focused app for analyzing blood work. It's similar to https://getbased.health, but simpler. Currently, you have to run the app locally and also have Ollama running on your machine. So there's more steps to get the app up and running, but the privacy guarantees are better and you can run the app completely offline once you have everything set up. The app gives you insights about specific lab panels as well as trends in your overall health and specific health markers. You can also chat about a specific lab panel or trends of a specific health marker. I've been using gemma4:26b-mlx with the app which works great on my Mac Mini M4 Pro with 64 GB of memory. I was originally using qwen3.6:27b which gave me good insights, but was really slow. gemma4:26b-mlx is a lot faster and generates responses similar to what I was getting from qwen. The repo includes a /compare-live-evals skill that can be used to help you choose what model works best on your machine. I currently only have live evals for panel chats though. I'm planning to add live evals for trend chats soon.\n\nhttps://blossom.primal.net/dd41187116efa35a4b19cdc585cc5ea33f1eb82b1dcacf1ca8409361ffe62c9b.png\n\nhttps://github.com/SamSamskies/locallab",
"sig": "8796254de55824bea97a272bde0258cbccee09c6ec624bb2b80a6fb12ebfb6c469a8010414cd9ba783e9ed935551da4c786cd6d21fbc3ff42d56695e2a00e93a"
}
