In the summer of 2030, wearable penetration in the US is 85% with over half of adults wearing one regularly. At-home blood testing makes it trivial to test blood monthly or even weekly. Continuous glucose and hormone monitors are mainstream.
Sensors are so cheap that millions of non-diabetic consumers try them. Biohacker subreddits and Discord groups discuss Biolinq’s new continuous multianalyte sensor that can track everything from ketones to electrolytes. Epigenetic testing is a common part of annual physicals and every baby gets their genome sequenced at birth.
The catalog of successful consumer passive tracking products continues to grow—Throne for toilet-integrated stool and urine analysis, AirPods for EEG, Eight Sleep for sleep tracking. Brain-computer interfaces are a proven tool for patients with neurological conditions while everyday consumers use tFUS devices to help with mood and focus.
The future of consumer health is exciting—implantable tooth sensors to track food intake, GI pills that measure gut health, and general purpose molecular sensing all have real paths to become hit consumer products in the next two years.
But healthcare has never just been about measurement and testing. Advances in hardware and testing make it easier to observe what is going on in patients’ bodies. The trend that matters, for the past decade and the next, is that the cost of producing causal evidence about oneself is falling. By orders of magnitude. Better, cheaper hardware, combined with smarter software, are driving down the cost curve that matters most for patients: how do I know if something worked for me?
Take blood glucose. Millions of people want to answer the question which foods are good or bad for my blood sugar? Ten years ago, running that experiment cost over $1,500 and required a doctor’s prescription:
Dexcom G5 receiver — $600
Dexcom G5 transmitter — $600
Dexcom G5 weekly sensor x4 — $300
Copay for doctor’s visit — $50
There were also no real consumer apps to track and analyze data. Five years later, the same experiment cost 75% less. The Dexcom G6 was a $400 transmitter, no receiver, and $300 for a month's worth of sensors. The Libre 2 got rid of the transmitter entirely and cut sensor cost in half. Patients still needed a doctor’s prescription, but there were now consumer apps that could track and interpret their data—and companies like Levels and Nutrisense bundled sensors and an app for a $400/month subscription.
Today, anyone can buy CGM sensors over the counter for less than $100/month. Tracking apps are abundant and ChatGPT can analyze your data. That same experiment is now 95% cheaper than it was a decade ago.
Over the next five years, Chinese manufacturers entering the US market will reduce hardware costs another 30-50%. Improvements in wear time will double the lives of most sensors. Apple and others are working on optical sensing which could bundle glucose monitoring into a wearable. A $400 Apple Watch could give multiple years of sensing. There is a path to reducing costs another 90%.
N-of-1
Why does all this matter? Medicine and healthcare today are built around population averages. The randomized controlled trial is humanity’s best solution to a set of constraints—it was expensive to segment patients, coordinate them at scale, and measure them continuously.
Those constraints are lifting: wearables and D2C testing provide rich profiles that can segment populations by biology and proactively find subjects. Software can help coordinate thousands of patients anywhere in the world. And continuous measurement removes the need for centralized and expensive trial sites.
The individual does not care about population averages if they can get an answer about themselves. Phase III data matters much less in driving patient decisions if the n-of-1 data is easily obtainable. As the cost of acquiring data about oneself continues falling, we believe consumers will allocate their dollars and time to things that are proven to be effective for themselves.
What does the world look like if the cost of n-of-1 evidence continues to get cheaper?
The test slot economy
Consumers can now get an answer to “does X work for me?” in weeks. Answers are cheap, but not yet too cheap to meter. People still need to run the experiment—take a drug or supplement, wait for it to wash in, measure for long enough to get a statistically meaningful result. Most experiments will take around 4-8+ weeks. Stacking interventions that target the same thing hurts attribution. Some interventions will also have washout periods.
In a year, the average person probably has 40-45 clean experiment weeks where they aren’t sick or traveling. At 6 weeks per experiment, that yields a maximum of ~7 per year. With optimized scheduling and stacking, that might increase to 10 or 12 per year.
The scarce resource is test slots. Therefore the most important question is: what should I test/experiment with in my next slot(s)? Whoever can best answer that question will direct a large chunk of the $150B+ Americans spend out of pocket on supplements and drugs every year. This is one of the questions that Ren aims to answer for consumers.
Luckily, there is valuable data from other people who have run experiments. The first shape of business solving for the test slot economy is a database of verdicts from thousands of users, filterable along axes such as bloodwork, genomic markers, lifestyle, and goals. Experimentation is a multiplayer product where people get guidance from others who have committed time to running experiments on themselves.
The test slot allocator starts as a pull product that users actively go to and search, but as the corpus builds and individual response measurement gets better, it becomes a push product. It reads fixed and stateful data, and proactively recommends experiments that have the highest expected value: “try magnesium right now because your training load is increasing and your recovery is getting worse; people with similar profiles saw 2x the population average effect” or “your biological aging test showed aging pace at 1.2x and slightly elevated HbA1c; ask your physician about metformin because users with similar metabolic signatures saw aging pace drop by 30%.”
Over the next five years, testing and measurement will continue to improve and the cost curve will continue to fall. The behavior of self-experimentation will become ubiquitous, and the asset we are building is longitudinal data overlaid on observational data across millions of users. Each experiment is a labeled piece of training data. We will begin to predict how different interventions will perform without running the actual experiment. The simulator will replace the wind tunnel and we will make the test slot economy obsolete by making experiments too cheap to meter.
Today, consumers search PubMed or Reddit to see if magnesium will improve their sleep. Tomorrow, Ren will give them the answer along with what other interventions are most likely to help. We will denominate healthcare in outcomes and make it abundant for everyone.
Thank you to Kyle, Vishal, and others for thoughtful feedback and discussion.