README / README.md
a2ed's picture
Update README.md
3021005 verified
|
Raw
History Blame Contribute Delete
1.66 kB
metadata
title: League Labs
emoji: πŸ§ͺ
colorFrom: blue
colorTo: green
sdk: static
pinned: false

League Labs

League Labs is the applied AI research group at League, building the intelligence layer for agentic healthcare β€” and publishing the datasets, benchmarks, and methodology openly.

League's platform powers benefits and care navigation for 70+ million people across payers and providers. A decade of building healthcare AI at that scale taught us what the field is missing: data and evaluation that reflect how people actually show up when they need care. League Labs exists to build that foundation and share it β€” openly, responsibly, and with the rigor healthcare demands.

What we work on

🧬 League SYNTH β€” Our platform for Privacy Preserving Personas: richly detailed synthetic patients and members, modeled from aggregate population-health statistics, that mirror the diversity and complexity of real populations without exposing a single record. SYNTH keeps the structured context that produced each sample β€” traits, situations, lineage, and label provenance ship with the data.

πŸ›‘οΈ Healthcare-native small language models β€” Purpose-built SLMs for agentic healthcare applications, designed to beat general-purpose models on cost, latency, and safety in member-facing workflows β€” starting with input guardrails that keep conversations safe, on-policy, and clinically appropriate in real time.

πŸ“Š Patient-centered benchmarks β€” Evaluations that measure AI against the real health journey: understanding options, navigating care, recognizing health signals in natural conversation, and knowing when to escalate to a human