BioMarkIQ
BioMarkIQ develops VarSee Lung, a blood-based, multi-protein panel combined with machine-learning specialist models for early-stage lung cancer detection. The platform measures a panel of protein biomarkers from a single blood draw and uses a gating model that weighs specialist readings alongside clinical context to produce a single result. The company is currently conducting a multicenter diagnostic accuracy study across three Tata Memorial Centre hospitals.
- Healthcare Technology
- Biotechnology
- Artificial Intelligence
- Digital Health
- Software Only
- Data & Analytics
Funding
Founders
Founder details are not available yet.
Product
Problem
Lung cancer remains one of the most prevalent and deadly diseases, and early detection significantly improves survival rates. However, the lack of a non-invasive, highly sensitive screening tool creates a formidable challenge in the healthcare landscape, often delaying diagnosis until later stages.
Solution
BioMarkIQ introduces VarSee Lung, a diagnostic platform that combines a blood-based protein panel with AI-driven analysis to enable early-stage lung cancer detection. The platform measures multiple protein biomarkers from a single blood draw and uses a mixture-of-experts (MoE) engine, where each specialist model reads the panel for one cancer type and a gating layer weighs their answers with the person's clinical context—age, sex, smoking, and recent blood count—to produce one result with a likely tissue of origin. When readings disagree, the output is indeterminate with a directed next step. The company is also working toward multi-cancer detection, with lung as the first program.
Target Audience
Primary customers include healthcare providers and clinical professionals seeking non-invasive cancer screening tools, with the current product available for research use. The company is also building toward multi-cancer detection for broader clinical application.
Features
- VarSee Lung measures a panel of protein biomarkers from a single blood draw, alongside routine clinical context such as age, sex, smoking, and recent blood count
- The MoE engine uses specialist models trained on proteins relevant to specific organs (lung, breast, liver, colorectal) plus a healthy-baseline specialist, with a gating model that weighs all readings in context
- Output provides a single result with likely tissue of origin, or an indeterminate result with a directed next step when specialist readings disagree
- For the multi-cancer panel, machine learning and scientific review narrowed approximately 2.5 million PubMed articles published since 2018 to 729 studies for detailed review and 18 candidate markers
- The platform is designed to optimize both sensitivity (finding cancer) and specificity (avoiding false alarms), with validation studies designed to demonstrate panel performance
- A multicenter diagnostic accuracy study has ethics approval at three Tata Memorial Centre hospitals, planned for about 1,000 participants