Demystifying cancer
Cancer is many diseases. Understanding what they share—and how they differ—is the starting point for looking earlier.
Read the storyScience notes
Understand the science
of early detection.
Biomarkers, cancer screening, and machine learning, explained by the BioMarkIQ team. Read the articles or explore how test results depend on the population.
START WITH THE QUESTION
Detect intelligently
Where machine learning can help cancer research, and why the biological question still comes first.
Read the story
Ten years of lung cancer screening—and what comes next
A screening program needs more than a test. Lessons from access, outreach, and the pathway after a result.
Read the storyCONTEXT CHANGES THE PICTURE
A percentage is
only the beginning.
Use hypothetical assumptions to explore sensitivity, specificity, and the population behind a result. This educational example is not VarSee performance.
EXPLORE AN EDUCATIONAL EXAMPLESynthetic assumptions · not VarSee results
People with the condition100
People without the condition9,900
Positive predictive value44.7%
A positive result’s meaning depends on the population as well as the assumed test properties. This is why context belongs beside a percentage.
Read definitions, calculations, and the data table
Expected counts use the hypothetical cohort and assumptions above. Values may be fractional; display rounding does not change the calculations.
| Outcome | Expected count | Calculation |
|---|---|---|
| True positives | 80 | N × prevalence × sensitivity |
| False negatives | 20 | N × prevalence × (1 − sensitivity) |
| True negatives | 9,801 | N × (1 − prevalence) × specificity |
| False positives | 99 | N × (1 − prevalence) × (1 − specificity) |
| Positive predictive value | 44.7% | True positives / all positive results |
| Negative predictive value | 99.8% | True negatives / all negative results |
Educational example only. This does not estimate your risk, model a VarSee test, or provide medical advice.