Mass General Brigham spinout Aevius launches with $22 million for patient-data AI
Aevius, a Boston health AI company spun out of Mass General Brigham, launched with $22 million led by Insight Partners to build models that turn a patient's scattered records into one computable picture.
Docket, covering health regulation and AI in care
, 3 min read
$22MAevius's launch funding, led by Insight Partners
What to know
- Aevius announced $22 million in funding led by Insight Partners on November 17; no other investors or round name were given.
- The company builds AI models that combine structured and unstructured medical data into a patient representation that updates as new data arrives.
- Aevius named no customers and published no performance figures.
Aevius, a Boston health AI company spun out of Mass General Brigham, launched on November 17 with $22 million in funding led by Insight Partners. The company's announcement names no other investors and does not call the round a seed or a Series A. It says the money will pay for model and product development, larger engineering, clinical and commercial teams, and commercial deployment.
What Aevius builds
Aevius says its AI models learn from structured, unstructured and specialty medical data held by hospitals, insurers and drug companies. HIT Consultant's report spells out what that covers: structured fields in the electronic health record (EHR), unstructured clinical notes, and specialty data such as pathology, imaging and genomics.
The product is what the company calls a refreshable multimodal patient representation: a single computable picture of a patient's clinical state and trajectory that updates as new information arrives. Several applications can be built on one representation, according to the company, including forecasting outcomes, finding clinically similar patients, sorting patients into populations and tracking change over time.
Aevius pitches this as a different approach from the chat-style AI tools now spreading through health systems. Co-founder and chief technology officer Ming Yang (Max) Lu says in the release that agents built on text-only language models assemble a patient picture one question at a time, while Aevius's models learn relationships across data types and across a patient's history. The company published no benchmarks or accuracy figures to back that comparison.
Who is behind it
The founders are Rowland Pettit, the chief executive, Lu, Faisal Mahmood and Long Phi Le. According to the release, they were part of the founding team behind Modella AI, which AstraZeneca acquired. Cardiologist and digital medicine researcher Eric Topol chairs the company's scientific advisory board.
Mass General Brigham's chief innovation officer, Chris Coburn, said in the release that "Aevius was spun out of MGB to bring together research developed here" with a team experienced in bringing medical AI to market. Neither the release nor HIT Consultant gives the terms of the spinout, a founding date or the size of the staff.
Who it is selling to
Aevius is aiming at three markets. For health systems, it lists clinical research, anticipating patient needs, and care delivery and resource planning. For insurers, it lists tracking patients' needs over time to support earlier, better-coordinated care. For drug and life sciences companies, it lists studying disease progression and outcomes under treatment, finding candidate biomarkers, and finding patients for clinical trials. The company named no customers in any of these groups.
The launch lands in a busy week for patient-level prediction models. A day earlier, UC Berkeley spinout Knit Health raised $35 million for Med-BEAM, a model trained on sequences of clinical decisions rather than text (our story). Health systems are also building their own companies: The Permanente Medical Group launched Permanentech this week (our story).
Why it matters
Most patient data sits in pieces: notes in one place, lab values in another, scans and pathology slides somewhere else. Aevius is betting that a single, continually updated model of each patient will be more useful to hospitals, insurers and trial sponsors than tools that read records on demand. Whether it is depends on evidence the company has not yet published. Prediction tools used to guide treatment can also fall under FDA device rules, depending on how they are used (our explainer), and claims about better outcomes are best tested in trials (our explainer).
This article is general information about how technology works. It is not medical advice. Talk to a clinician about your own health.
Sources
- Aevius Launches with $22 Million in Funding to Make Medicine Computable, Aevius via Business Wire (BioSpace), November 17, 2026
- Mass General Brigham Spinout Aevius Raises $22M to Make Medicine Computable, HIT Consultant, November 17, 2026
Digital health and AI in careAI models
Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.