OM1 turns the clinical paper trail into a view of what comes next

The revealing number is 650,000. That is how many women, at minimum, appeared in the real-world study behind an expanded screening indication for Hologic’s Aptima HPV Assay. These were records from multiple U.S. health systems. Somewhere inside that enormous collection of ordinary care was evidence a regulator could use. The difficulty was making it legible.

• The business: clinical data, AI predictions and automated evidence programs for healthcare organizations.
• The distinction: specialist detail and measured outcomes, rather than a warehouse of billing codes alone.
• The practical test: validated records that can support research, treatment conversations and regulatory review.

Hologic announced FDA approval for clinician-collected primary HPV screening in February 2026. In May, OM1 described its contribution: automated acquisition, quality control, reporting and extraction from unstructured clinical documents. The models used to read those documents were validated and documented for regulatory review. The unglamorous requirement was traceability. Every useful answer needed a defensible path back through the data.

The bill tells you surprisingly little

A medical bill can establish that a visit occurred, a procedure happened or a prescription was filled. Researchers often want something more awkward: whether a patient improved, how severe the disease became, why treatment changed. Those details may live in a specialist’s note, a clinical scale or the patient’s own assessment. Healthcare is exceptionally good at generating paper trails. Making those trails comparable is another occupation.

OM1 occupies that space. Founded by physician Richard Gliklich in 2015, it gathers and organizes longitudinal clinical information, builds disease-specific datasets and applies AI to measure and predict outcomes. Its mission joins three jobs: accelerating research, measuring what happens to patients and improving clinical decisions. None is accomplished merely by accumulating another million rows.

More records, or better records?

Gliklich had already founded Outcome, a research technology and services company acquired in 2011. OM1 followed after time in venture capital. In a 2022 interview, he described a consequential adjustment: the company initially aggregated data from many sources, then recognized that specialist networks could supply deeper clinical information. Chronic disease care has its own vocabulary. A rheumatologist’s measures and a psychiatrist’s observations contain different kinds of evidence.

That is the useful business lesson here. Begin with the question the data must answer, then pursue the people who actually record the answer. The shift was toward clinical depth. It helps explain why OM1’s offerings span dermatology, rheumatology, mental health, neuroscience and other therapeutic areas, with expertise in epidemiology and outcomes research alongside engineering.

“Clinical outcomes are the most important metric in healthcare.”Richard Gliklich, 2019 financing announcement

A fingerprint with a clinical purpose

PhenOM, introduced in 2023, is OM1’s healthcare foundation model. The company reports training it on more than one billion patient-years. A patient-year measures time represented in medical histories; it does not mean one billion separate patients. OM1’s data cloud separately reports coverage of more than 350 million patients. The two figures describe different things.

The model uses patterns in records to create digital phenotypes, sometimes described as patient fingerprints. A research team can look for likely undiagnosed patients, distinguish subgroups or estimate future medical events. Delivery can take the form of an API, application, EHR integration or report. OM1’s published workflow explicitly includes validation between analysis and delivery. That middle step deserves attention: a plausible prediction still needs a performance check.

Consider hidradenitis suppurativa, a painful inflammatory skin condition that can go unrecognized for years. The American Academy of Dermatology launched a Novartis-supported project in October 2024 combining DataDerm’s more than 63 million deidentified encounters with OM1’s Patient Finder. By September 2026, Novartis described work translating the analysis into physician resources. Finding missed patterns becomes useful when someone can recognize them during care.

The prediction enters the consultation

Joint Insights offers a more intimate example. The AI-enabled decision aid combines education, a patient’s preferences and personalized outcome reports for knee osteoarthritis. A 2021 randomized trial examined decision quality, patient experience and functional outcomes. A later randomized trial, published in eClinicalMedicine in 2025, tested an AI-enabled aid against education alone at UT Austin’s Musculoskeletal Institute.

The attraction is easy to grasp. Someone considering knee replacement wants to understand likely outcomes in their own circumstances. A personalized report can give patient and clinician something concrete to discuss. The later study was single-site and open-label, however. That setting bounds what readers should infer about performance elsewhere. The sensible application is an informed conversation, with predictions interpreted in their clinical context.

Evidence is the product; infrastructure is the expense

OM1 sells to life sciences companies, device makers, providers, payers and medical associations. Buyers can license curated cohorts, commission scientific work, automate registries or use prediction tools. Aspen, launched in June 2023, supports automated studies. The wider platform connects data, extracts information from notes and supplies quality dashboards, with clinicians and data scientists reviewing workflows.

The model is a mix of data licensing, technology and research services. Its financing reveals the scale of the build: a $21 million Series B in 2018, a $50 million Series C in 2019 and an $85 million round in 2021. The last was led by D1 Capital Partners, Kaiser Permanente and Breyer Capital to expand the data cloud and registries. Those are capital raised, rather than a customer’s study budget.

OM1 sits among evidence businesses with different specialties: Aetion in evidence analytics, Flatiron in oncology data, IQVIA across data and research services. Its particular pitch combines specialist clinical depth, outcome measurement and automation. The lesson travels beyond medicine: define the decision, preserve the detail and validate the answer. The Hologic example makes that sequence tangible. A vast medical archive became evidence for a specific screening question. That is a considerably more useful ambition than simply having a vast archive.