We Built an Adoption Index for Precision Medicine. Here Is What Eight Technologies Told Us.
Everyone in this industry has an opinion about where a technology sits on the adoption curve. Almost no one measures it. We decided to try — to take eight precision medicine technologies and place each one, year by year, somewhere between Innovators and Laggards, using nothing but the published literature. The result is the Adoption Index. The more interesting story is what it took to build it.
The leading-indicator trap
The starting premise was simple and, it turned out, half-wrong. When a technology is genuinely new, it punches above its weight in the most competitive journals — Nature, Science, Cell. As it becomes routine, that novelty premium erodes. So the ratio of top-tier papers to total papers should tell you where a technology sits on the curve. Clean, quantitative, defensible.
Except it lied. The novelty premium is a leading indicator — it collapses years before adoption actually peaks. When we let it drive the phase assignment, the formula confidently declared single-cell RNA-seq, spatial transcriptomics, and even multi-omics to be in Late Majority, because their Nature-cover days were behind them. Meanwhile those fields were publishing more papers every year than the year before. A technology cannot be in Late Majority while its adoption is still accelerating. The premium had faded, but the mainstream had not even arrived.
That is the trap. The signal that is easiest to measure — are the elite journals still excited? — answers a different question than the one you are asking. The elite get bored long before the market matures.
Teaching the formula what we already knew
The fix was to add a second, concurrent signal: the growth rate of annual publications itself. Accelerating from a small base means a field is still pre-chasm. Decelerating but positive means the majority is piling in. Growth at or below the ambient rate of science — roughly four percent a year, the background hum of PubMed expanding — means the field has become standard practice. Pairing the leading signal (premium) with the concurrent one (growth) is what makes the index work.
But here is the part worth being honest about, because it is the part that makes the index credible rather than suspect: the thresholds were not handed down by the math. We set them by hand, calibrating against what we already knew to be true. We knew NGS was a commodity. We knew long-read was a real but maturing platform. We knew high-plex multi-omics was just getting started. So we tuned the dials — the growth rate that marks Late Majority, the publication volume that signals mass adoption regardless of journal mix — until the framework reproduced reality, and then we froze them and let the index run.
For a while the formula kept insisting that things were Laggards when our gut said they were nowhere near it. The breakthrough was not a cleverer equation. It was accepting that a tiny field is Innovators by definition — you cannot be late to a party of thirty people — and that absolute scale is a floor on how far along you can possibly be, no matter what the noisy journal ratios say. Once we encoded the qualitative truths as guardrails, the quantitative signals had room to be useful. That is the whole method in one sentence: quant for the trajectory, qual for the boundaries.
The NGS single-year chasm
The first thing the calibrated index revealed is that NGS crossed the chasm in a single year. It sat in Early Adopters for 2008 and was in Early Majority by 2009 — a transition that took single-cell RNA-seq five years and has taken spatial proteomics six and counting.
This is not noise. NGS arrived into pent-up demand with an obvious job to do. Sanger sequencing had already defined the goal — sequence everything, cheaply — so the moment Illumina made it real, the mainstream did not need years of "is this even useful?" exploration. It skipped the application-discovery phase because the application was already universal. The lesson generalizes: a compressed Early Adopters phase is the signature of a technology meeting demand that was already there. When the market is waiting, the chasm is narrow.
The aggregate hides the frontier
The second lesson is a caution about reading the index too literally. cfDNA registers as Late Majority. Taken at face value, that suggests a mature, finished space. It is nothing of the sort.
A single phase per technology is a weighted average across many sub-applications, and cfDNA is really three markets at once. Therapy selection is genuinely late-stage — Guardant, Foundation, routine and reimbursed. Minimal residual disease monitoring is squarely in Early Majority, expanding fast into solid tumors. And multi-cancer early detection is still in the Innovators phase, with the defining evidence yet to read out. The aggregate tells you where the center of mass is. It does not tell you where the opportunity is. The opportunity is almost always in the sub-segment the average is hiding.
The hypothesized commercial window
The most actionable thing the index produces is not a phase at all. It is the gap between the two signals.
When the novelty premium has collapsed to the floor but publication volume is still climbing — the index flags this on NGS and single-cell RNA-seq — a technology is in what we came to call the hypothesized commercial window. The science risk is gone. The elite journals have moved on precisely because the thing works and is no longer surprising. But the market is still expanding. That combination — proven, unglamorous, growing — is historically the most favorable moment to build and sell a product. It is the opposite of the Innovators phase, where the excitement is maximal and the addressable market is a rounding error. The index does not tell you what is exciting. It tells you what is safe and still growing, which for anyone allocating capital is the more useful signal.
What it is for
The usual proxies for technology maturity — conference buzz, venture funding, analyst coverage — are all lagging. By the time a technology is on the cover of a review journal, the premium phase is over. Publications are earlier, and the ratio of where they land to how many there are is a real-time readout of how the scientific community is treating a technique. The eight curves in the Adoption Index are not forecasts. They are measurements of something the market is usually a few years behind in pricing.
It is also, deliberately, a living instrument. Re-run it next year and the phases move. That is the point. We built it to be a recurring read on the industry — a KPI for where the tools of precision medicine actually are, as opposed to where the press releases say they are. The companion benchmark analysis has the full taxonomy. This note is about how the thing was made, and the one lesson that mattered most: a model is only as good as the reality you are willing to teach it.