Work
None of this was commissioned. Each piece started as a question nobody had answered properly, and each one is live, in use, and checkable. That is a stronger reference than a logo wall, because you can test every claim below yourself.
Could public data find a base site before the operator did?
A county-level demand and siting model for air medical transport, built entirely from public sources: the federal provider registry, census data, CMS claims and state registries. Every base located and attributed to its operator, every county scored for demand, payer mix and distance to definitive care.
The model flagged Colby, Kansas as uncovered demand. A base was publicly announced there afterwards. The model did not know; it only had the data.
What it proves: a model built from data anyone can download can see what an operator with inside knowledge sees, and say so first. About Meridian
What happens when the test says your own finding is wrong?
An earlier version of the siting work produced a spatial-clustering result that would have made a persuasive slide: bases appeared to cluster in a way that implied a strategy. A permutation test, run before anyone was shown it, said the pattern was no stronger than chance.
The result was withdrawn and the withdrawal was published alongside the work. Separately, the published site-ranking models the market sells were tested the same way and did not predict outcomes either, p = 0.86. That was published too, rather than selling a competing one.
What it proves: the standard holds when it costs something. Anyone can say evidence first; this is a case where the evidence said no and the answer was still no.
What does demand look like when you stop modeling it and go get it?
Modelled demand is an estimate. Minnesota keeps an actual air medical transport registry. A state public records request produced it: 49,074 transports over five years, by county. Measured, not inferred. Five further state requests are outstanding.
In parallel, published in-network air ambulance rates were extracted from ten payers' public price-transparency files across six states: 270 rates covering 17 organizations, and a verified finding that one national payer holds no air ambulance contract at all in four of the six states. That is revenue per transport by payer and geography, which is what a break-even actually needs.
What it proves: the hard part is usually getting the real number, not building a cleverer estimate of it. The pipelines that parse those terabyte-scale payer files exist because a terabyte is not something anyone reads by hand.
Can an operating standard be written so that it can be audited?
A 34-standard operating playbook across eight pillars, from aviation and clinical through dispatch, billing, human resources and business development. Every standard carries the evidence behind it and its own KPI block. Every pillar carries an implementation sequence. Nothing in it needs a capital budget to start.
It was written from the operating seat by someone who had run a 2.4 million dollar service back to profitability and cut its overtime by 30 percent, which is why it says what to do first when everything cannot be done at once.
What it proves: "operational excellence" can be made into numbers an organization can be held to, and sequenced so it starts tomorrow. The playbook
Does the method work on a problem at your own kitchen table?
The founder's son reads accurately and slowly: 40 words correct per minute on unseen text, against a grade-level norm of 83, at 93 to 95 percent accuracy. The tools on offer did not look at what he was actually getting wrong. So the method was applied at home: the national oral reading fluency norms read from the technical report, the intervention literature on dosage, the mastery criterion research, and a tool built to match.
It records every error against the pattern underneath it, feeds the weak patterns back into the next week's practice, and refuses to advance a level until performance holds at 90 percent across three separate days. Over a thousand automated tests sit behind it. It is in daily use.
What it proves: the method is not an industry trick. Research, build, measure works on a reading problem the same way it works on a market, and the person who built it had to live with the result. About Able Minds Lab
Next
The first conversation is to find out whether the method fits. If it does not, we will say so and point you somewhere better.
