Alzheimer's in 2026: blood tests, AI brain scans and the new drugs explained

For the first time, Alzheimer's can be confirmed with a blood test and slowed with medicine. The gains are real but modest, and access depends heavily on where you live.

What has changed

Until recently, Alzheimer's was diagnosed mostly by ruling out other causes, and the drugs available, such as donepezil or memantine, only eased symptoms for a while. Confirming the underlying biology required a lumbar puncture or an expensive PET scan.

Three developments have changed that picture: blood tests that detect the disease's hallmark proteins, AI software that reads routine brain scans, and antibody drugs that remove amyloid plaques from the brain.

Blood tests: how they work and how accurate they are

In Alzheimer's, a protein called amyloid builds up into plaques, and another protein, tau, becomes chemically altered. Fragments of altered tau, especially one called p-tau217, leak into the blood. Some tests also measure the ratio of two forms of amyloid (Aβ42/Aβ40).

In 2024 the Alzheimer's Association added these blood markers to its official diagnostic criteria, alongside PET and spinal fluid tests.

The key evidence came from a study led by Lund University in Sweden, published in JAMA in July 2024. It tested about 1,200 people with memory complaints. The blood test was right about 90% of the time in both primary care and memory clinics. By comparison, family doctors relying on usual assessments were right 63% of the time, and dementia specialists 73%. A small share of people, roughly 6% to 11%, fell into a grey zone and needed a confirmatory scan or spinal tap.

AI that reads brain scans

In January 2024 the FDA authorised BrainSee by Darmiyan, software that analyses a standard brain MRI together with two memory test scores. For people aged 55 and over with mild memory impairment, it estimates the chance of progressing to Alzheimer's dementia within five years. Other AI tools under study analyse digital clock-drawing tests and retinal images.

The new drugs: lecanemab and donanemab

Both are antibodies given by infusion that help the immune system clear amyloid from the brain.

• Lecanemab (Leqembi, Eisai and Biogen): an infusion every two weeks. In the 1,795-person Clarity AD trial, it slowed decline on the main rating scale by 27% over 18 months.
• Donanemab (Kisunla, Eli Lilly): an infusion every four weeks. In the 1,736-person TRAILBLAZER-ALZ 2 trial, it slowed decline by about 35% in people with lower tau levels. A distinctive feature is that treatment can stop once scans show the plaques have cleared, which happened in about three quarters of patients by 18 months.

In practical terms, this is roughly four to six extra months at a given level of ability over a year and a half. Benefits were larger when treatment started at the mild cognitive impairment stage. Some experts argue the effect is below what patients would notice day to day; others point to growing benefit over longer follow-up.

Who can be treated

• People with mild cognitive impairment or mild dementia due to Alzheimer's. The drugs are not used in moderate or severe disease.
• Amyloid must be confirmed by PET, spinal fluid or an approved blood test pathway.
• APOE genetic testing is needed. In the EU and UK, people with two copies of APOE4 are excluded because their risk of brain swelling is much higher.
• European labels advise against use in people taking blood thinners.

The main risk: ARIA

ARIA stands for amyloid-related imaging abnormalities: swelling (ARIA-E) or small bleeds (ARIA-H) in the brain as amyloid is cleared. Most cases cause no symptoms and are seen only on MRI, so regular scans are scheduled in the first months. When symptoms occur they can include headache, confusion, dizziness or vision changes. Rarely, ARIA has been fatal. Risk is highest in people with two APOE4 copies, at over 30% for swelling in trials.

Why the US and Europe decided differently

The FDA approved both drugs based on statistically significant slowing and plaque removal. European regulators first rejected both, then approved them only for lower-risk patients. In England, NICE accepted that the drugs work but concluded the benefit was too small for the cost to the NHS, which includes infusions and repeated MRI scans. Germany launched donanemab commercially in September 2025, with reimbursement decided by national bodies.

What did not work: GLP-1 drugs

Hopes that diabetes and weight-loss drugs might protect the brain were tested in two large trials of oral semaglutide (EVOKE and EVOKE Plus, about 3,800 people). In November 2025 Novo Nordisk reported that the drug did not slow Alzheimer's progression.

What is coming next

• Trontinemab (Roche): uses a "brain shuttle" to carry the antibody into the brain more efficiently. Early studies showed fast plaque clearance with fewer cases of swelling; phase 3 trials are under way.
• Valiltramiprosate (Alzheon): a tablet tested specifically in people with two APOE4 copies.
• Subcutaneous lecanemab: an injection pen version that could allow maintenance treatment at home.
• Anti-tau antibodies such as bepranemab, aimed at the second protein in the disease.

What to ask your doctor

• Would a blood test help clarify what is causing my memory problems?
• Am I at an early enough stage to be considered for treatment?
• What is my APOE status, and how does it change my risk?
• Where would infusions and MRI checks happen, and who pays?

Questions patients ask