What Is Your Health Data Worth? Prices From Real Deals

Every figure in this article comes from CureCircle's evidence base: a published deal, a filing, a guideline or a pay rate, quoted word for word and re-checked at its source every day. Prices move, so the date of each source is given with it. Read on 6 November 2026.

Your health data is being bought and sold already, just not by you. Pharmaceutical companies pay hundreds of millions of dollars for access to patient records, genetic databases change hands in bankruptcy auctions, and wearable companies are valued by how many people they track. The people the data describes almost never see any of that money.

So what is one person's health data actually worth? The honest answer is a range, and it depends on four things you can measure: what kind of data it is, how much of it there is, how rare the condition behind it is, and how many kinds of data are linked together for the same person. This guide works through each one with the real prices behind it, and ends with what a typical contribution comes to.

The short answer

A complete account of one condition, together with a year of the records patients with that condition usually have, is worth roughly $40 to $300 per licence to a research buyer, on CureCircle's benchmark built from the prices below. A common condition such as type 2 diabetes sits near $91; a rare one such as multiple sclerosis near $275, because research on rare conditions pays more per patient and finds fewer of them.

Two things matter about that number. It is per licence, and health data is licensed non-exclusively: the same de-identified record can be licensed to several research teams. And it is a benchmark, not a promise. Nobody pays for a single person's record on its own; buyers license datasets, and a record earns when a dataset that includes it is licensed. You can value your own data with the same benchmark, line by line.

What one person's data has sold for

The clearest evidence of what health data is worth comes from deals where the price and the number of people are both public, so the price per person can be worked out by simple division. There are fewer of these than you might think, and each comes with caveats, but together they set the scale.

Patient records licensed for AI drug discovery. In its 2025 annual report, Tempus states that Pathos will pay it data license fees of $200 million over three years, and that Tempus holds more than 8.6 million de-identified patient records. Divided out, that is about $23 per record. It is an illustration rather than an exact price, because the licence may not cover the whole database, but it is one of the largest disclosed licences of patient records to an AI company, and it is the anchor of the benchmark's value for one patient record.

Genetic data, sold in bankruptcy. When 23andMe collapsed, Regeneron bought the company for $256 million, covering roughly 15 million people who had taken its test. That works out to about $17 per person, and it is a floor: it was a distressed sale, and the price also bought the brand, the lab and the biobank. Before the collapse, 23andMe was licensing the same database for real money: GSK paid a $20 million upfront fee for a one-year, non-exclusive data licence in 2023.

Exomes, licensed to drug companies. In 2018, Regeneron formed a consortium in which each of five companies paid $10 million for early access to the exome sequences of 500,000 UK Biobank volunteers. Per company, that is $20 per exome, and because five companies paid, the same exome earned five times over. That is the non-exclusive licence model in one deal.

Insurance claims, in bulk. Claims data is the cheapest health data per person because it is collected automatically and in enormous volume. When PRA Health Sciences bought Symphony Health for $530 million, Symphony's data covered more than 280 million lives: about $1.89 per person.

Wearables. Google agreed to buy Fitbit for about $2.1 billion, when Fitbit had 28 million active users: about $75 per active user. Like the 23andMe figure, this is a whole-company price that includes a hardware business, so it says more about how investors value a tracked population than about the price of a day of heart-rate data.

Health records combined with genomes. The consultancy EY estimated, from market transactions and company valuations, that electronic health record data is typically worth under £100 per patient record, but £1,000 to £5,000 per record when combined with genomic and phenotypic data. That gap, a factor of 10 to 50, is the single most important fact about health data value, and we come back to it below.

What research buyers pay per year

Most health data is not sold per person at all. Buyers license access to a dataset for a year at a time. A 2022 review in Clinical Pharmacology & Therapeutics gives the going rates in the United States: private payer claims data generally costs $100,000 to $300,000 per therapeutic area per year, structured EHR data $1 to 3 million, and specialty unstructured EHR data that needs extensive curation $3 to 5 million.

The same database can carry very different prices depending on who buys it and for what. Recursion's agreement with Tempus, disclosed in an SEC filing, commits it to $160 million over five years, with annual licence fees between $22 million and $42 million. Public research datasets sit at the other end: the US government's national inpatient sample, covering millions of hospital stays, lists at $750 for a year of data.

The lesson for an individual is that data from one person only becomes valuable inside a dataset, and a dataset is valued by what it lets a buyer learn. That is why the next section matters more than any single price.

What makes health data worth more

If you compare the prices above, the same few factors explain almost all of the differences. They are also the factors you can influence when you decide what to share.

Linked data: the biggest multiplier

A blood test result on its own tells a researcher very little. The same result linked to the person's diagnosis, their medication history, their scans and how they responded to treatment is a different thing entirely. The market prices that difference explicitly. In the claims and EHR rates above, curated EHR data costs 10 to 50 times as much as claims data per therapeutic area, because it links more of a patient's story. EY's estimate shows the same factor of 10 to 50 between a health record alone and one combined with genomic and phenotypic data.

CureCircle's benchmark takes the most conservative version of this: the richness premium is the lowest linkage ratio found in the market, 3 times, earned in full when a contribution links four or more kinds of data (your own account, lab results, letters, scans and so on), and in part when it links fewer.

Rarity: research on rare conditions pays more per patient

A rare condition has fewer patients to study, and every one of them matters more to a trial. A study in the Orphanet Journal of Rare Diseases found that clinical trial costs per patient for orphan drugs are about 2.5 times those for non-orphan drugs. What counts as rare is set in law: in the European Union, a condition affecting not more than five in 10,000 people.

The benchmark uses the low end of the published cost ratios as its rarity premium, about 2.9 times typical, applied to conditions under that threshold. Whether a condition is rare is worked out from its own published prevalence: the number of people living with it worldwide divided by the world's population, each from its source, shown on every condition page.

Time and depth: a series beats a snapshot

Researchers learn the most from change over time: how a marker moved before and after a treatment, how a condition progressed over years. The benchmark counts this through quantity: a repeated test is valued per result, with a diminishing return on each additional one (quantity to the power 0.8), so a long series is worth more than a single result but not linearly more. Guidelines tell us what a typical series looks like. For type 2 diabetes, NICE's guideline calls for an HbA1c test every 3 to 6 months until stable, then every 6 months, which is why a diabetes contribution usually carries a lab series.

The kind of data

Some kinds of data are simply priced higher per unit than others, because they cost more to produce and answer more questions. A whole exome licensed at $20 per company is worth far more than a single lab result. Where the market prices a kind of data directly, the benchmark uses that price; where it does not, it values the kind as a stated share of one patient record, and says so on the methodology page.

What patients are paid directly

There is a second market that is easy to overlook: research that pays patients directly for their experience. It is the closest thing that exists today to a price for a person's own health story, and it is far higher per hour than most people expect.

Market research firms recruit patients for interviews and surveys about their condition, and the rates are published. A list of current Rare Patient Voice studies shows $50 for a 30-minute survey with rheumatoid arthritis patients, $120 for a 60-minute interview with metastatic breast cancer patients, and $210 for a 90-minute interview with Crohn's and ulcerative colitis patients. The CLL Society advertised $400 for a 90-minute interview plus a short journal for people with chronic lymphocytic leukaemia. Rare Patient Voice itself says it pays patients and caregivers around $120 per hour. Every rate by condition, from surveys to trials, is in our guide to what patients are paid for research.

There is even a published standard for it. The US National Health Council's fair market value methodology, which sponsors use to set what they pay patients for advisory and engagement work, puts an individual patient or caregiver at $55 to $120 per hour, and an experienced patient expert higher. CureCircle uses these rates as the value of a patient's time spent sharing: about $100 an hour, shown beside every contribution's value so you can see both.

Sensor data has its own going rate in research studies. A University of Southern California study pays $35 for wearing a Fitbit through a study of about three weeks, and a type 1 diabetes study at the Joslin Diabetes Center pays $75 for each collection of continuous glucose monitor data. Worked out per day, that puts a day of wearable data at about $1 and a day of glucose sensor data at a few dollars.

Putting it together: a worked example

Here is how the benchmark values a complete contribution from someone with type 2 diabetes who shares their story and the records that guidelines say people with type 2 diabetes usually have. Every line links to its derivation on the type 2 diabetes page.

Type 2 diabetes is common, so there is no rarity premium. The same contribution for a rare condition would carry one: for multiple sclerosis the benchmark comes to about $275 per licence, almost all of the difference coming from rarity. The story itself is the largest single line, because a patient's own account of how a condition started, what was tried and what worked is exactly what records leave out.

The catch: why your data alone is worth little, and together worth a lot

It is worth being clear about what these numbers do not mean.

Nobody buys one person's record. Every price above is a share of a dataset. One record earns when a dataset that includes it is licensed, and how often that happens depends on how many buyers want data on that condition. That is why CureCircle shows a value per licence rather than a lump sum.

People value their data far above market prices. When researchers asked German health app and wearable users what they would want to be paid for their data, the average answer was €237.30 a month from pharmaceutical and device companies and €145.66 a month from universities. That is many times what any deal above pays per person per year. Part of that gap is because people's data is usually taken without being asked, so asking feels like the start of a negotiation. Part of it is real: data a person chooses to share, complete with their own account, is worth more than data scraped from a billing system.

Most of the value is captured upstream today. In every deal above, the money went to a company holding the data, a lab, an insurer, a hospital or a platform, and none of it to the people described. The prices show the value exists. Who receives it is a choice.

How to get paid for yours

CureCircle is built on that choice. You share your health story in a short conversation, add the records you have (lab results, letters, scan reports, even a diary you kept), and keep your own organised medical file. Research partners license de-identified datasets, and you earn a share each time a licence includes your data. You can withdraw at any time.

Every value we show is computed from the evidence in this article and drills down to its sources. If you want to see what yours could be worth, value your health data or browse the conditions to see the figures for yours. And if you want to check our working, the full methodology and every source are public.