We Can Now Detect Alzheimer's a Decade Early With a Blood Test. We Still Can't Do Much About It. That Gap Is a Business Problem.

The most important development in dementia this year is not a drug. It is a blood test. At the Alzheimer's Association International Conference in London this month, researchers reported that a blood test measuring a protein called p-tau217 helped family doctors in Sweden diagnose Alzheimer's correctly 93% of the time, nearly matching specialists' 94%, and that the same class of test may flag risk five to ten years before any symptoms appear.

That is a genuine breakthrough, and it creates a genuine problem. Medicine can now detect Alzheimer's cheaply, early, and in an ordinary doctor's office, well before it can do much to change the disease's course. The gap between what can be diagnosed and what can be treated is widening, and who pays to build the infrastructure that lives inside that gap is the question the conference was quietly circling.

The diagnosis-treatment gap, stated plainly

For most of medical history, diagnosis and treatment advanced together, because there was little reason to detect a disease you could not act on. Alzheimer's has now split those two things apart.

On the diagnosis side, the progress is real and fast. Blood-based biomarkers like p-tau217 are moving Alzheimer's detection out of specialist centers and expensive PET scans and into primary care and routine blood draws. A family doctor can increasingly identify the disease, and a predictive version can flag people at risk a decade before symptoms, which is exactly the window in which intervention would theoretically do the most good.

On the treatment side, the progress is real but modest. The first disease-modifying therapies have been approved, drugs that clear amyloid from the brain, and they do slow decline. But they slow it by a limited amount, they carry meaningful risks including brain swelling and bleeding, they require infusions and monitoring, and they are expensive. They are a genuine first step and they are not a cure, and for someone flagged as at-risk a decade early, there is currently no proven treatment to offer at all.

So the field has arrived at an uncomfortable place: it can increasingly tell you that you have Alzheimer's, or will, and it cannot yet do very much about it. That is not a reason to dismiss the diagnostic advance, since knowing changes planning, clinical-trial enrollment, and the management of treatable contributors. But it is the structural fact that shapes everything about the business model.

Why detection without strong treatment is a hard thing to fund

Here is the problem the conference framed as a business-model question rather than a scientific one. Building the infrastructure to screen an aging population for Alzheimer's, the tests, the workflows, the trained clinicians, the systems to act on results, costs real money, and someone has to pay for it. What that infrastructure produces, right now, is mostly diagnoses that do not lead to a dramatically better outcome.

In a normal medical market, you screen because early detection lets you intervene and change the course of the disease, and the value of the intervention justifies the cost of the screening. Cancer screening works this way: you fund the mammogram because catching the tumor early means you can treat it and save a life. Alzheimer's screening does not yet fit that model cleanly, because catching the disease early does not yet lead to a treatment that changes the trajectory enough to obviously justify the system-wide cost of finding it.

That creates a genuine tension for the health systems and companies that would build the infrastructure. A hospital deciding whether to standardize cognitive screening across its population, or a company selling the tools to do it, is investing in a capability whose payoff depends on treatments that mostly do not exist yet. The bet is that the treatments are coming, and that the systems which build the detection infrastructure now will be positioned to capture the value when they arrive. That is a plausible bet. It is also a bet, made ahead of the evidence, on the timing of scientific progress nobody controls.

The case for building it anyway

There is a real argument that the infrastructure is worth building even before the treatments justify it, and it deserves to be stated fully rather than treated as mere optimism.

First, detection has value independent of disease-modifying drugs. A person who knows early can plan their life, finances, and care while they still have the capacity to make those choices, can enroll in clinical trials that require early-stage patients, and can address the treatable things that worsen cognitive decline, cardiovascular risk, medication interactions, sleep, that are manageable now. Diagnosis is not nothing even when a cure is absent, and for many families the certainty itself has value.

Second, the infrastructure is a precondition for the treatments working when they arrive. Disease-modifying drugs appear to work best early, before too much damage is done. If effective treatments come, the health systems that can already detect the disease early will be the ones able to deploy them, and building that capability from scratch after the fact would waste years. Detection infrastructure is the rails the future treatments will run on.

Third, prevention research needs it. Much of the most promising work is on risk reduction, the lifestyle and vascular interventions that may delay or prevent decline, and you cannot run those programs or measure their effect without the ability to identify at-risk people early. The screening capability is the foundation of the prevention agenda, not just the treatment one.

All three are real, and together they make a serious case. They also all depend, to varying degrees, on future developments rather than present ones, which is exactly why it is a business-model question and not a settled one.

The ethical weight the business framing understates

There is a dimension the pure business analysis leaves out, and it deserves naming because it is the human core of the issue. Telling someone they will likely develop an incurable dementia in ten years is a profound act with real consequences, and doing it at population scale raises questions the infrastructure conversation can gloss over.

Early diagnosis can bring the benefits above, and it can also bring anxiety, depression, insurance and employment discrimination, and years of living under a diagnosis for which nothing can yet be done. Different people will weigh that tradeoff differently, and many will reasonably choose to know. But a system built to screen broadly has to reckon with the fact that a predictive result, absent a treatment, hands people knowledge that some will find empowering and others devastating, and that the same result can be either depending on the person and the support around them. The value of detection is not uniform, and a responsible screening system has to be built around informed choice rather than default testing, which adds cost and complexity to the very infrastructure whose economics are already uncertain.

How to read it

The clean way to hold this is that dementia care has entered a strange and consequential in-between period. The tools to find the disease early have arrived faster than the tools to treat it, and that mismatch is not a scientific failure but a predictable stage in how a field matures, detection often outruns cure. The open question is who invests in the capability to act on early detection during the years before the treatments catch up, and on what expectation of return.

For health systems and the companies serving them, the strategic bet is whether to build cognitive-health infrastructure now, on the expectation that better treatments are coming and that early positioning will pay off, or to wait until the treatments justify the cost and risk being years behind when they do. There are honest arguments on both sides, and the conference framing suggests the field increasingly believes the build-now case, though belief is not the same as certainty about timing.

For everyone else, the useful thing to understand is the shape of the moment. A blood test that predicts Alzheimer's a decade out is a remarkable achievement and a genuine good, and it lands in a world that does not yet know what to do with the answer at scale. Closing that gap, turning early detection into early, effective intervention, is the real work ahead, and until it is done, the infrastructure being debated is a bet on a future that has not arrived, built by people who believe it is coming. The science made it possible to know. The harder question, the one the business model is really about, is what knowing is worth before there is something to do about it.

Primary sources

1. The Alzheimer's Association International Conference 2026 press materials for the Swedish real-world study of more than 1,300 patients and 165 physicians showing family doctors diagnosed Alzheimer's correctly 93% of the time versus 94% for specialists, and the p-tau217 predictive-risk findings five to ten years before symptoms.
2. Clinical Trials Arena and Pharmaceutical Technology for the AAIC 2026 discussion of blood-based biomarkers bridging primary and specialty care and the arrival of the first disease-modifying therapies and blood-based tests.
3. Becker's coverage via Creyos for health systems standardizing cognitive screening and the framing that organizations investing in cognitive-health infrastructure now will be positioned as the aging population's needs grow.
4. Alzheimer's Research UK for the Dementia Frontiers Fund and the emphasis on earlier detection alongside prevention.
5. BioSpace for the Alzheimer's Association's Brain Health Advancement Institute and its dementia-risk-reduction public-health framing.