Cognito’s Alzheimer’s headset fails to slow decline in trial


Spectris AD missed the main goal of a yearlong trial against a sham device. Cognito plans to seek US authorization, while independent specialists say findings in a milder subgroup need further study.
Cognito Therapeutics’ experimental Spectris AD headset failed to slow decline compared with a sham device in a trial of people with mild-to-moderate Alzheimer’s disease. The company said the study missed its main goal across the overall group.
Cognito presented the findings at the Clinical Trials on Alzheimer’s Disease meeting in Boston on 16 November, Nature said. An earlier, smaller trial published in 2024 also failed to slow cognitive and functional decline. Called OVERTURE, it involved 76 participants over six months.
Spectris AD combines goggles and headphones that produce flashing light and rhythmic sounds. They are designed to induce gamma waves, a type of brain activity linked to attention and memory. Cognito says the stimulation targets 40 Hz.
The multicenter HOPE trial randomly assigned 673 participants to use an active or sham device at home for 60 minutes a day over 12 months. The study was double-blind: neither participants nor researchers knew who received the active treatment during the test.
The main measure, the integrated Alzheimer’s Disease MMSE and ADCS-ADL Rating Scale (iADMARS), combined equally weighted scores for global cognition (MMSE) and daily functioning (ADCS-ADL). In the primary analysis population (mITT-a) of 663 participants, Cognito said the active and sham groups did not differ in change from baseline on that measure at month 12; the company also found no statistically significant differences on the trial’s secondary measures across the overall group.
| Group | Average decline |
|---|---|
| Active device | 8.97 points |
| Sham device |

Nature gave those average changes for the two groups after the year of treatment.
Analysis
Cognito chief executive Christian Howell said the device showed benefit among participants with milder impairment. Nature said active-device users in that subgroup declined 41.6% more slowly than sham users, an average difference of 3 points.
Cognito had planned the subgroup analysis in advance. It included 395 participants who scored 21–28 at baseline on the Mini-Mental State Examination (MMSE). Participants were randomized within mild (MMSE 21–28) and moderate (MMSE 15–20) strata. On the trial’s main combined measure, the company calculated 42% slower decline in the subgroup, with a p-value of 0.072. A p-value describes how unusual a result would be under a statistical assumption of no difference; the company said its subgroup p-values were not adjusted for testing multiple comparisons.
Eric McDade, a Washington University School of Medicine neurologist who was not involved in the study, called the overall trial negative and recommended a follow-up study in the milder group. Rob Howard, a University College London psychiatrist specializing in dementia, said the subgroup data were exploratory rather than persuasive proof that the device works.
Cognito said adverse device effects occurred in 18.3% of the 333 participants in the active group and 15.5% of the 330 in the sham group. It said there were no unanticipated adverse device effects in either group.
Spectris AD is not authorized by the US Food and Drug Administration, and its safety and effectiveness have not been established, Cognito said. Howell told Nature the company plans to file for approval in 2027. Cognito intends to seek authorization for people with early symptomatic Alzheimer’s disease. Spectris AD has received the FDA’s Breakthrough Device designation, Cognito said. Cognito also plans to discuss the evidence and requirements for a De Novo authorization submission with the regulator and continue collecting longer-term data in an extension in which participants know they are receiving the device.
Researchers found that 80% of chatbot switches to incorrect climate answers occurred at the first mild user disagreement. Chatbots were mostly accurate about individual climate policies, but performed no better than random guessing when asked how two policies would work together.
1 min

The tally reported as of 15 November 2026 uses data through 14 November. Community transmission continues, while ECDC retains its low risk estimate for EU/EEA travelers in affected areas and very low estimate for people living in the EU/EEA.
4 min

The journal withdrew the Denali Therapeutics study after a coauthor found that Madhuja Samaddar had changed data points, affecting some figures and conclusions, the retraction notice says.
2 min