Xlue Launches a Health World Model With a $35 Million Seed to Predict Disease Risk and Simulate Clinical Trials

Xlue, a Carnegie Mellon spinout, launched with a $35 million seed round led by Conviction. It builds a health world model that learns how a patient's health state changes over time, aiming at disease risk prediction and simulated clinical trials.

Christopher Ort
By i10X News Desk · Reviewed by Christopher Ort · 3 min read
Share
A physician seen from behind reviews a long printed patient history timeline in a hospital office.
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. Background
  5. Why it matters
  6. What's still open
  7. About the analysis
  8. i10x Perspective

Xlue, a startup spun out of Carnegie Mellon University, has launched with a $35 million seed round led by Conviction. The company is building what it calls a human health world model: an AI that learns how a patient’s health state changes over time, with the goal of predicting disease risk and simulating clinical trials before they are run.

i10X Quick Take

Quick Take

  1. The money: A $35 million seed round led by Conviction, the AI investment firm founded by Sarah Guo.
  2. The model: A foundation model built around health states and treatment actions rather than text tokens.
  3. The claim: Xlue says the model outperforms frontier token-based models in blind third-party evaluations at less than 0.1% of the compute cost.
  4. The evidence so far: In an October evaluation run by data company Protege on 2,500 held-out patients, Xlue’s model flagged the most future cases on 14 of 16 disease risk prediction tasks, at about $0.04 per 1,000 predictions.
  5. The first targets: Prospective disease risk studies with health systems and simulated clinical trials with drug developers.

In plain English

Chatbots know a lot of medical facts, but they cannot tell you when a particular person is likely to get sick. Xlue’s model learns from many patients’ records over time, such as diagnoses, treatments and test results, and tries to forecast where a specific patient’s health is heading. The same idea could let drug companies simulate how groups of patients might respond to a treatment before running an expensive trial.

What happened

Xlue went live on November 18 with a technical post and announced its seed round on X. CEO Chenyan Xiong, a Carnegie Mellon professor, leads the founding team.

In its launch post, the company writes that today’s most advanced AI models “demonstrate impressive medical knowledge, but they cannot predict when our loved ones fall sick, nor reason about which treatment will work.” Its answer is to model “how human health states evolve over time, and how actions change their trajectories.” Xlue says it is “conducting prospective studies of disease risk prediction across multiple health systems” and working with pharmaceutical companies to find new indications for existing drugs.

Background

Xlue is not alone in pitching world models for medicine. Earlier this week Knit Health unveiled Med-BEAM, which it describes as a world model for healthcare trained on longitudinal patient records, alongside a $35 million Series A led by NEA. Drug developers are also adding AI to trial operations, as with Oracle’s new life sciences agents for clinical trials.

Why it matters

Most medical AI products so far summarize notes or answer questions. Predicting who will get sick, and simulating how patients respond to a drug, would move AI into the decisions that drive outcomes and costs. Retrospective scores are only a first step, though: prospective studies will show whether earlier flags lead to earlier and better care.

What’s still open

  • Results from the prospective studies Xlue says it is running with health systems.
  • How often the model raises false alarms, and what that means for extra testing.
  • Which regulatory path applies if the model is used to guide clinical decisions.

About the analysis

This independent i10x analysis is based on Xlue’s launch post, its funding announcement and the Protege evaluation published on Xlue’s website. Quotes are taken from the launch post.

Sources (plain text, no links):

  • Xlue: “Human Health World Models for Real-World Clinical Intelligence” (November 18, 2026)
  • Xlue announcement on X of its $35 million seed round led by Conviction (November 18, 2026)
  • Knit Health newsroom: “Building a World Model for Healthcare” (November 2026)
  • Protege: “Independent Evaluation of Xlue’s Disease Risk Prediction Model” (October 21, 2026)

i10x Perspective

If language models were the medical student, Xlue wants to build the resident who has seen thousands of patient journeys. The pitch is strong; the proof will come from prospective data. Track the field in our State of AI in Health & Medical Research report and on the Startups page.

Christopher Ort

Christopher Ort

Christopher Ort is the editor of i10X News at i10x.ai. He has worked hands-on with AI tools, models and agent workflows for years and leads i10X’s coverage of AI launches, funding and research. He is the author of “The Style Penalty”, an SSRN study on how frontier AI models judge AI-written CVs, covered by heise online and t3n.

Editor's picks

  1. Dragos canary study: attackers stole AI prompts and credentials at three AI inference providers
  2. Analysis: AI Agents in Banking Hit a Liability Wall
  3. CVE-2026-133631: FlowIntel chatbot let LLM call any API, exposed Ollama token in process table
  4. Stripe Evals: Coding Agents Score 78% on Payment Integrations, but Only 29% Pass Every Required Check
  5. Kong Volcano: A Self-Healing Agent Cloud Where Coding Agents Build, Host and Fix Apps