Hepta

Market position

Hepta is an early-stage, seed-funded private company positioning itself in the gap between invasive liver biopsy, scarce specialized imaging, and existing non-invasive blood scores such as FIB-4 for MASH diagnosis and treatment selection. It frames MASH as the largest chronic disease diagnostic gap in the US and argues that the arrival of approved MASH therapies increases the need for an accurate blood-based triage test. Third-party datasets classify it as angel/seed stage in biotechnology, healthcare IT and medical AI, and list comparable companies including AGED Diagnostics, Glycotest and EarlyDiagnostics.

Hepta describes LiquidTransformer as the first liquid biopsy-native AI platform built specifically for the faint molecular signal regime of chronic disease, in contrast to cancer-oriented liquid biopsy methods that look for rare, distinct mutations, copy-number variation, or unique epigenetic changes. Technical differentiators cited are single-molecule attention-based weighting of cfDNA fragments, simultaneous processing of very large numbers of molecular interactions rather than segment-by-segment genome analysis, and a patented model architecture. Clinically, the company points to a reported AUC of 0.86 for significant MASH fibrosis with a threefold reduction in false positives versus existing blood tests, validation across multiple biopsy-confirmed cohorts, and validation of blood signals against paired liver tissue methylation, gene expression, and single-cell multi-omics.

Technology

LiquidTransformer is a patented transformer-based model that reads epigenetic (methylation) patterns in cell-free DNA from plasma. It uses attention to determine the significance of individual cfDNA fragments at single-molecule level and, according to the CTO, processes all molecular interactions simultaneously rather than analyzing the genome in smaller segments, in order to detect subtle methylation shifts distributed across millions of genomic sites — the signal regime characteristic of chronic disease rather than cancer. Outputs are intended to reflect the molecular state of the source tissue, enabling inference of liver pathway activation, inflammation and fibrosis from blood. Validation rests on the MASH Atlas, an ultra-deep paired liver-tissue and plasma cfDNA methylation sequencing study combined with gene expression and single-cell multi-omics, plus multiple biopsy-confirmed cohorts.

Go-to-market

The company's near-term go-to-market is evidence generation and clinical validation: peer-reviewed conference presentations (EASL 2025, AASLD The Liver Meeting 2025, MASH-TAG 2026, EASL 2026, DDW 2026), manuscripts in submission or preparation, and collaborations with academic hepatology centers and a large scientific advisory board of hepatology key opinion leaders. Seed funding is designated for dataset expansion, platform scaling, regulatory milestones, and commercialization to bring the assay into clinical use. The company also solicits collaborations with clinicians, researchers, and partners via its website.

Clinicians and healthcare providers managing patients with suspected MASH or MASLD, including primary and non-tertiary care settings where liver biopsy and specialized imaging are unavailable; clinical laboratories and healthcare institutions as purchasers; and, indirectly, drug developers and researchers, given the therapy-response and mechanism-of-action applications and clinical-trial diagnostics vertical.

Geography

Headquarters are reported in the San Francisco Bay Area, California (San Francisco per several sources; Preqin lists 'California City, CA'). Clinical and academic collaborators span the United States (Duke, UCSD, Mayo Clinic, Mount Sinai, Cleveland Clinic, UCSF, University of Chicago, Virginia Commonwealth, Saint Louis University, Houston Methodist) and Europe and Asia (Saarland University in Germany, Newcastle University in the UK, Sorbonne University in France, KTH in Sweden, and the Chinese University of Hong Kong). One secondary source reports an intention to expand into international markets in Europe and Asia in coming years.

History

Sources give conflicting founding years: Preqin and GenomeWeb state 2022, while Caplight and Startup Intros state 2021. The company operated in stealth before announcing its $6.7 million seed round and public launch on 5 November 2025 (one aggregator instead says it emerged from stealth in 2023). Around the launch it published a multi-omic MASH atlas (10 November 2025) and presented the paired liver-tissue/plasma cfDNA methylation atlas at AASLD The Liver Meeting 2025. Subsequent milestones listed on its site include EASL Congress 2025 and 2026 presentations, MASH-TAG 2026, DDW 2026, and a 5 May 2026 release on epigenetic signatures of GLP-1 response.

Risks & controversies

No controversies are reported in the sources. Risk-relevant observations: the company is pre-commercial with a single disclosed seed round of $6.7 million and a headcount reported at six; regulatory pathways and commercialization are stated goals rather than achievements; key clinical claims (AUC 0.86, threefold false-positive reduction, GLP-1 response prediction) originate from company statements with manuscripts still in submission or preparation rather than published peer-reviewed papers. Source records also conflict on basic firmographics — founding year (2021 vs. 2022), stealth exit year, headquarters designation (San Francisco vs. 'California City, CA'), and employee count (6 vs. an implausible '10000+' listing) — indicating limited third-party verification.