BioMarkIQ
CompanyPartner with us
Menu
BioMarkIQEvidence
Evidence

How we evaluate
VarSee Lung.

Early detection asks two things of a test: find the cancers that are there, and avoid raising alarms when they are not. We study those questions together.

01SENSITIVITY

Find the signal.

Among people who have the disease, how many does the approach identify? Disease stage matters: an early-stage cancer can present a different detection challenge.

02SPECIFICITY

Separate the noise.

Among people who do not have the disease, how many does the approach correctly identify? Avoiding false positives matters for the people who would otherwise face further investigation.

03FROM SIGNAL TO EVIDENCE

Every stage
asks a new question.

A test earns trust one question at a time: does the signal exist, does the model hold on new people, and does it hold across hospitals and stages. Here is where VarSee Lung stands.

  1. Which markers carry the signal?

    What it measures
    Evidence from the literature, curated by scientists and scored for early-stage performance and biology.
    VarSee Lung
    About 2.5 million articles screened and 729 studies curated in depth to choose the panel.

  2. Can a model read the pattern?

    What it measures
    A model trained on a development cohort, then locked: its inputs, weights and threshold fixed.
    VarSee Lung
    The lung model was developed on a pilot cohort.

  3. Does it hold on new people, with nothing changed?

    What it measures
    Sensitivity and specificity against a histopathology reference, in people the model has never seen.
    VarSee Lung
    Evaluated independently at Tata Memorial Hospital, with no change to the model or its threshold. The evidence base covers 424 people across three cohorts.

  4. How sensitive at each stage, and how specific, across hospitals?

    What it measures
    Stage-wise sensitivity and specificity in the intended population, at several sites.
    VarSee Lung
    Ethics committees at Tata Memorial Centre hospitals in Mumbai, Varanasi and Sangrur cleared the diagnostic accuracy study, planned for about 1,000 participants.

  5. Does finding it earlier change what happens next?

    What it measures
    Stage at diagnosis, treatment and outcomes in practice, and the regulatory approval that clinical use requires.
    VarSee Lung
    Available for research use only today. Not for diagnostic use.

24.2%of low-dose CT screens were positive over three rounds of the US National Lung Screening Trial.

96.4%of those positive results were false positives.

At population scale, false alarms add up quickly. That is why we study specificity as closely as sensitivity.

NLST Research Team, N Engl J Med, 2011 (opens in a new tab)

Test the panel.
Test the contribution.

We look for complementary information, then ask what it adds. A useful comparison studies the panel and the individual-marker baseline in the same setting.

Independent evaluation asks the next question: does the pattern hold beyond the data used to develop it?

Explore complementary signals

Keep each
disease in view.

A combined multi-cancer result can hide differences between cancers and stages. Our bottom-up approach keeps those individual questions visible as we work toward broader coverage.

Research collaborations help define the population, the comparison, and the study needed to answer them.

Discuss a research collaboration
FURTHER READINGNational Cancer Institute · Cancer screening (opens in a new tab)

Find more cancers. Raise fewer false alarms.

Let’s design the right study.

Discuss a partnership