We paid a doctor to find problems. Here is what she found.

Most health products that say "medically reviewed" never tell you what the reviewer actually said. This page does. We hired an independent physician to audit every evidence claim in the tool, told her in writing that the review was allowed to fail, and published her findings in full, including the four things we have not fixed.

Who reviewed it

Susana Huacuja, MD

Reviewed the Solenna tool on 19 August 2026. Corrections verified 28 August 2026. Photograph published with her permission.

Susana Huacuja, MD. A physician licensed in both Colombia and Mexico, working as a population health coordinator for a US healthcare organisation, with daily clinical contact with an elderly patient population that includes people living with dementia. Her registration was verified in Colombia's ReTHUS register and Mexico's SEP professional register before anything on this page was published.

She was paid for the review. We are saying so plainly because a paid reviewer who is not disclosed is a testimonial pretending to be an audit. What she was not paid for was approval: the contract stated that if she concluded a feature should not exist in its current form, she would be paid in full and the finding published.

What she reviewed, in her words

"I reviewed the risk profile, brain workout, and dashboard screens against the evidence claims and disclaimer language as of August 19, 2026. I did not review the blog, article library, or marketing copy. This reflects the product as it existed on that date, not an ongoing endorsement."

— Susana Huacuja, MD

What she found, and what we changed

A privacy claim that was false

The result screen said "your answers never leave this device" directly above the optional email field, which sends your address and your ranked levers to a server. It was the single most serious finding in the review: a false statement about data, placed exactly where a reader relies on it. It now states precisely what is sent and what stays on the device.

An inflated blood-pressure statistic

We had written that blood-pressure control "alone cut mild cognitive impairment by 19%". She identified four problems: 19% is a relative reduction and the absolute figure was 1.4 percentage points over five years; it was a secondary endpoint in a trial whose primary endpoint was not significant; "alone" misdescribed a comparison in which both groups received medication; and the effect weakened with longer follow-up. Rewritten to her wording.

A superlative our own source does not support

We called physical activity "the most consistently protective behaviour there is". The Lancet Commission, which the tool cites as its authority, puts physical inactivity at roughly a third of the population share of hearing loss, calls the relationship likely bidirectional, and never ranks it first. Corrected.

A misread of our own index

We stated that depression carries the single highest weight in the LIBRA index. It carries the largest weight among the risk-increasing factors; a protective factor carries a larger one. Corrected.

A population estimate described as personal risk

"About 45% of dementia risk" is not what that figure means. It refers to dementia cases, across a whole population, that could theoretically be prevented or delayed if all fourteen factors were eliminated. It has never meant that 45% of any individual's risk sits in those factors. Rewritten, including the Commission's own caveat that it cannot guarantee any individual avoids dementia.

Trials that did not exist

The dashboard referred to "large 2025 trials", plural, linking blood pressure to fewer dementia cases. There is one such trial. The other 2025 papers are an earlier trial's follow-up and a second study, both null. Now singular, described, and dated.

The brain-training claim, which was the reason we hired her

The tool weights its drills toward processing speed on the strength of the ACTIVE trial, and four places on the site said the product was "grounded in", "following", "anchored on" or "built on" it. ACTIVE was ten supervised sessions of an hour or more over five to six weeks, using one specific software programme, with booster sessions later, and dementia was never a pre-specified outcome. At five years the same team found no effect. At twenty years, with registry-linked outcomes, the main effect disappeared.

Her conclusion was blunter than our own concern had been: an indefinite daily app is not more of ACTIVE's protocol, it is a different protocol with no evidence of its own. Every reference is now "informed by", and the dashboard says outright that no brain-training programme has been shown to prevent dementia.

An undisclosed clinical instrument

Our mood question is a close paraphrase of a two-item depression screen used in clinical practice, shortened and rescaled. It now says what it is adapted from, that it is not that instrument, and that mood concerns belong with a doctor.

A hearing claim that was too broad, and a missing warning

The best trial of treating hearing loss found no effect across everyone it enrolled and a real effect in the group already at higher risk. The screen now says both halves. Separately, the Brain Score trend reassured people about normal day-to-day variation but never addressed what a genuine sustained decline is not. It now does, and the persistent score card carries its own reminder that the number is a benchmark for those drills and not a diagnosis.

What we have not fixed

These are open. They are on this page because leaving them out would make the rest of it dishonest, and because a reviewer's name should never imply she cleared something she did not.

• The scoring weights are out of date. The index we score against was updated in 2024 with recalculated weights and three additional factors. We have not migrated to it.
• One factor has no basis in that index. Vision was never part of it in either version, and the weight we assign is unanchored.
• Two different factor counts. The questionnaire scores fourteen items; the dashboard grid checks eleven. They are not the same fourteen, and reconciling them properly is a design decision we have not yet made. In the meantime the screen states plainly which set it is using and which of the Commission's factors it does not ask about.
• The regulatory question. Her assessment is that tracking a cognitive score over time, and bundling it with blood-pressure readings into a document intended for a doctor, sits close to the line European and US regulators draw between a wellness tool and a medical device. That is a decision about what the product does, not a wording fix, and it is unresolved.

Why publish this

Because "medically reviewed" as a badge is worth very little. Anyone can buy one, and the reviews that produce them are usually not allowed to fail. A review that found a false privacy claim, an inflated statistic, a superlative contradicted by our own source, and a research citation that does not support the feature it justifies is worth something, and it is only worth something if you can read it.

If you find something we have got wrong, tell us. That is how this page got written.

Susana Huacuja, MD was paid a fee for this review. She reviewed the tool for accuracy and safety; she does not endorse outcomes for any individual user, and this is not an ongoing endorsement. Solenna provides educational, risk-reduction information and is not a medical diagnosis, treatment, or device. It does not prevent, treat, or cure Alzheimer's disease or any form of dementia. Always consult a qualified healthcare professional about your personal health.