Exploring the frontiers of medical AI integrity.

Exploring the frontiers of medical AI integrity.

Exploring the frontiers of medical AI integrity.

Our mission is to shape the future of trustworthy AI in healthcare and life-science.

Through rigorous evaluation and transparent data science, Lumos research helps ensure AI systems can reason like clinicians, learn from real patients, and improve outcomes without compromising ethics or safety.

Research

Research

Dec 24, 2025

Automatic Replication of LLM Mistakes in Medical Conversations

Distillation pipeline to turn errors from multi-turn conversations into zero-shot vignettes for SFT bank

Dec 24, 2025

Automatic Replication of LLM Mistakes in Medical Conversations

Distillation pipeline to turn errors from multi-turn conversations into zero-shot vignettes for SFT bank

Read paper

Dec 24, 2025

Automatic Replication of LLM Mistakes in Medical Conversations

Distillation pipeline to turn errors from multi-turn conversations into zero-shot vignettes for SFT bank

Read paper

Dec 24, 2025

Automatic Replication of LLM Mistakes in Medical Conversations

Distillation pipeline to turn errors from multi-turn conversations into zero-shot vignettes for SFT bank

Dec 21, 2025

A Women's Health Benchmark for LLMs

The first benchmark for evaluating LLM performance in women’s health

Dec 21, 2025

A Women's Health Benchmark for LLMs

The first benchmark for evaluating LLM performance in women’s health

Read paper

Leaderboard

Dec 21, 2025

A Women's Health Benchmark for LLMs

The first benchmark for evaluating LLM performance in women’s health

Read paper

Leaderboard

Dec 21, 2025

A Women's Health Benchmark for LLMs

The first benchmark for evaluating LLM performance in women’s health

Oct 13, 2025

Medical interpretability and knowledge maps of LLMs

Identifying the best layers of post training

Oct 13, 2025

Medical interpretability and knowledge maps of LLMs

Identifying the best layers of post training

Read paper

Oct 13, 2025

Medical interpretability and knowledge maps of LLMs

Identifying the best layers of post training

Read paper

Oct 13, 2025

Medical interpretability and knowledge maps of LLMs

Identifying the best layers of post training

Sep 29, 2025

MedPI eval: interaction-first medical AI evaluation

The first real multi-turn evaluation framework

Sep 29, 2025

MedPI eval: interaction-first medical AI evaluation

The first real multi-turn evaluation framework

Read whitepaper

Read technical paper

Leaderboard

Sep 29, 2025

MedPI eval: interaction-first medical AI evaluation

The first real multi-turn evaluation framework

Read whitepaper

Read technical paper

Leaderboard

Sep 29, 2025

MedPI eval: interaction-first medical AI evaluation

The first real multi-turn evaluation framework