Open model beats Jev on clinical decisions
pplx-decider v1.1 scored 643/669 on clinical decision tests, outperforming Jev's 628, with 42% lower cost per decision and similar speed. Weights are downloadable for on-premise hospital deployment.
THE JEV GUIDE · CURATED EDITION
Healthcare.
Launch news, explanations, demos, and community projects around TypeSafe Jev, organized by topic. Every entry links back to its X source.
Curated selection · Not exhaustive · Updated manually
5419curated posts
2789original videos
Sponsors
These products pay for the directory. Want yours up here?
JEV Powers Cross-Country Medicine Search
Mr Vu shares a sneak peek of a cross-country medicine search powered by JEV for search planning and EmbeddingGemma 2 for embeddings.
Jev and pplx-decider handle 669 clinical decisions for about 3 cents
This post shares cost and speed data for Jev and other models on clinical decision tasks, comparing total cost across 9 APIs and noting the fastest 3GB local model responds in 284 ms.
Hospital Triage Demo: Jev Picks Next Patient
In the demo, 20 patients wait with 3 doctors and 3 rooms. When a room frees up, Jev (TypeSafe Choice) picks the next patient based on sickest ESI level first, with ties broken by earlier arrival.
JevRad: An Educational Radiology Tool Built on Jev
A radiology resident built JevRad (educational use only) using TypeSafe's Jev model. After describing a patient, Jev picks the ACR scenario in 0.5s and selects the most appropriate imaging studies. Link
One card in the feed, every sixteen, on every page of it. $50 a week, cancel any week.
Mores Research Uses TypeSafe's Jev for Pediatric Mental Health Behavioral Intent Mapping
Udit Akhouri announced Mores Research, a lab using behavioral classifiers such as TypeSafe AI's Jev to conduct long-horizon real-time behavioral intent mapping for pediatric mental health.
Jev scores 16/16 on medical diagnostic test
OpenMed tested on four fictional medical notes: Jev scored 16/16, Laya 13/16; on medication-conflict questions 4/4 vs 2/4. The author notes this is a useful diagnostic, not a clinical accuracy estimate.
InvisibleBench Built with Jev Showcased at Caregiving Conference
Ali Madad presented givecareapp's InvisibleBench, rebuilt with TypeSafe AI's Jev, at a caregiving research conference in Utah.
Jev reads CT scans
Jev model demonstrates reading CT scans, potentially for medical imaging analysis.
Community Test: Jev Performs Strongly on Medical QA
A user tested Jev on the medical QA benchmark igakuQA: 2,000 questions solved in about 80 seconds for $0.04; text-only accuracy was 88.6%, comparable to GPT-4o; using confidence cutoffs further improved accuracy.
MedJev, inspired by Jev: structuring clinical notes on consumer GPUs
Inspired by Jev and the open-source community, MedJev turns free-text clinical notes into structured fields and runs on consumer GPUs.
Jev and LLMs in the Clinic: Choosing, Scoring, Routing
The post explains that Jev is not another chatbot but is built to choose, score and route, while an LLM drafts, summarizes and converses. In a clinic, the LLM handles language, Jev checks or routes the output, and staff stays in control.
