The CureCircle valuation model
How we value your health data
Nobody publishes a price per patient for health data, so we built one from what is published: what patients are paid for research interviews, what AI training work pays, and what research buyers paid in disclosed health data deals. Every story is valued two ways.
1. Your time
minutes / 60 x hourly rate
Valued at what paid patient research interviews pay: 60 to 100 USD an hour, and 100 to 150 for rare conditions.
2. Your data, per licence
base x rarity x completeness x documents x diagnosis x outcome
What one de-identified story is worth to a research buyer each time a dataset holding it is licensed. You earn a share of every licence.
What stories are worth
Computed by the model, as a contributor would see them.
A common condition, told in full, no documents
$34
per licence, $22 to $76
$27
time value of 20 minutes, $20 to $33
A rare condition, with lab results attached
$137
per licence, $42 to $531
$52
time value of 26 minutes, $43 to $65
An ultra rare condition, still ongoing, with records
$182
per licence, $58 to $630
$60
time value of 30 minutes, $50 to $75
Every parameter, and where it comes from
Measured: a published rate or price. Derived: our arithmetic on a disclosed deal. Assumption: our judgement where no price is published yet, replaced by real licence prices as they come in.
Hourly rate for a patient's time
Measured
60 to 100 (80)USD per hour
Moderated consumer research sessions most often pay 60 to 100 USD an hour (25,000+ sessions, 2025). Unmoderated sessions pay about 40, and the floor for any research participant is 8 to 12, so this is the rate of a guided interview, not of a survey.
Hourly rate, rare conditions
Measured
100 to 150 (120)USD per hour
A rare-disease patient panel pays 100 USD an hour (2017) and 120 for new studies since 2023; flat fees of 200 to 400 USD for a 60 to 90 minute rare-condition interview match it.
Base value of one patient record, per licence
Derived
17 to 60 (27)USD per record
No data vendor publishes a per-patient price, so this is derived from disclosed deals: 256 million USD for about 15 million 23andMe customers (17 each, a distressed sale), 200 million USD for a 3-year licence of Tempus's 7.3 million patient dataset (27 each), and GSK's 300 million USD with 23andMe for about 5 million customers (60 each).
Rarity: rare condition
Measured
1.5 to 3.5 (2.5)multiplier
Per patient, orphan-disease trials cost about 2.5 times non-orphan trials, because the patients are scarce. A record of a rare condition replaces patients a buyer would otherwise struggle to find.
Rarity: very rare condition
Assumption
2 to 4.5 (3)multiplier
Extends the measured 2.5 for rare conditions as patients get scarcer. No published figure separates very rare from rare.
Rarity: ultra rare condition
Assumption
2.5 to 5 (4)multiplier
Capped at 5, the top of the plausible band in the research. No published figure for ultra rare conditions.
Completeness of the story
Assumption
0.4 to 10.4 (basics only) to 1 (complete)
A record missing parts of the journey or treatment ratings is worth less: the factor runs from 0.4 (only the basics) to 1.0 (every part of the story covered and every treatment rated). The treatment ratings are the outcome labels a buyer pays for.
Clinical documents attached
Assumption
1.3 to 2 (1.6)multiplier
Licensed datasets step up in price with clinical depth: claims data about 0.1 to 0.8 million USD a year, structured health records 1 to 3 million, curated clinical records 3 to 5 million. A story with lab results or letters attached moves up that ladder; we apply a conservative 1.3 to 2 times.
- PMC9299990, real-world data licensing costs, about 2022
Confirmed diagnosis
Assumption
1.1multiplier
A doctor's diagnosis makes the record comparable with clinical data.
Known outcome
Assumption
1.15multiplier
A story that reached an outcome (recovered, improving, under control) says what worked, which is the scarcest thing in health data.
How rare is rare
- CommonMore than 5 in 10,000 people
- Rare1 to 5 in 10,000 people
- Very rare1 in 100,000 to 1 in 10,000 people
- Ultra rareFewer than 1 in 100,000 people
What health knowledge pays elsewhere
- Generalist AI training and annotation8 to 35 USD an hourDataAnnotation, Outlier and Prolific published rates, 2026
- Clinicians training AI (nurses, pharmacists, specialists)60 to 180 USD an hourMercor clinician network listing, 2026
- Physicians training AI (US)100 to 200 USD an hourHandshake AI physician listing, 2026
- Patients in paid research interviews60 to 120 USD an hourUser Interviews and Rare Patient Voice, 2023 to 2025
What we do not know yet
No data vendor publishes what it pays or charges per patient, and nobody has priced patient-told stories before. So the base value comes from disclosed deals, and the effect of documents, a confirmed diagnosis and a known outcome are our judgement. As research partners license CureCircle stories, their real prices replace these parameters, and this page shows the change.