Alzheimer's and brain decline
Diseases such as Alzheimer's destroy brain cells, and no treatment stops or reverses the decline.
Open in the interactive tree →Alzheimer's, Parkinson's, ALS and related disorders involve misfolded proteins and progressive loss of neurons, usually starting years before symptoms. The first antibody drugs clear amyloid plaques and slow decline only modestly, with a risk of brain swelling and bleeding. The chain of causes is still debated.
As of October 2026
WHO estimates 57 million people lived with dementia in 2021, with nearly 10 million new cases a year and a cost of 1.3 trillion US dollars (2019). The FDA gave full approval to the amyloid antibody lecanemab in July 2023 and approved donanemab in July 2024 (EU approvals in April and September 2025); brain swelling or bleeding occurred in 36.8 percent of donanemab patients versus 14.9 percent on placebo, and a Cochrane review of 17 trials with 20,342 participants (April 2026) concluded that these antibodies probably make little to no difference to cognition. On 16 May 2025 the FDA cleared the first blood test to help diagnose Alzheimer's (Lumipulse pTau217 to beta-amyloid ratio), whose negative results correctly ruled out amyloid pathology in 97.3 percent of cases.
What is missing
- A settled causal model: the roles of amyloid, tau, inflammation, blood vessels and infection
- Treatments that stop tau spreading or protect neurons, not only clear plaques
- Ways to deliver drugs across the blood-brain barrier
- Years-long prevention trials in people with biomarkers but no symptoms
- Cheap, accurate biomarkers for early detection and tracking
Becomes possible once solved
- Prevention of dementia in people at risk
- Stopping Parkinson's disease and ALS
- A longer, independent old age
Open steps
- A settled causal model Medium AI leverageDecide the roles of amyloid, tau, inflammation, blood vessels and infection in neuron loss.
- Drugs that protect neurons High AI leverageMolecules that stop tau or alpha-synuclein spreading or protect neurons, not only clear plaques.
- Crossing the blood-brain barrier Medium AI leverageWays to deliver large drugs such as antibodies into the brain at useful doses and safely.
- Cheap early biomarkers Medium AI leverageCheap, accurate biomarkers for early detection and tracking before symptoms appear.
- Prevention trials before symptoms Low AI leverageTrials that follow people with biomarkers but no symptoms for years to show that treatment delays decline.
Where AI could help
Medium AI leverage. AI speeds drug and biomarker discovery, but unclear causes, the blood-brain barrier and years-long prevention trials set the pace.
- Screening and generating molecules that block protein aggregation or protect neurons
- Finding disease subtypes and targets in brain-tissue, genetic and imaging data
- Combining blood, imaging and cognitive data into early-detection and progression markers
- Designing smaller, better-targeted trials by predicting who will decline
Shown so far
- In April 2024 Nature Chemical Biology published a Cambridge machine-learning screen of millions of compounds that found five potent blockers of alpha-synuclein clumping in Parkinson's disease, with screening about ten times faster (according to the university). source
Prerequisites
- Neuron doctrine1888
- Alzheimer's disease described1906The plaques and tangles Alzheimer described define the leading dementia
- L-DOPA for Parkinson's1967Parkinson's, a neurodegenerative disease, got its first effective treatment from L-DOPA
- Monoclonal antibodies1975The first disease-modifying Alzheimer's drugs are antibodies
- Functional MRI1992
- Single-cell atlases2016
- Alzheimer's antibody drugs2023The first drugs that slow decline came from targeting amyloid