55 APOE4 carriers used red light therapy for 4months: This is what happened
We ran our own study and tracked memory, sleep, and bloodwork in carriers in real life situations.
Key takeaways · TL;DR
We ran our own study and tracked memory, sleep, and bloodwork in carriers in real life situations.
Hi Phoenix friend,
Here’s a question your doctor probably can’t answer yet.
The big Alzheimer’s-prevention studies happen in tightly controlled trials. Selected people. Fixed doses. Clinic-grade machines. A short, funded window. They’re good at answering one thing: can this work?
But you don’t live in a trial. You live in a kitchen, a job, a family, a calendar that doesn’t care about your protocol. So the real question, the one that actually matters to an APOE4 carrier, is harder:
Does any of this hold up in a real life?
We decided to find out. Not in a lab. In the lives of real carriers, tracked continuously for four months. This is what we found.
If you’d rather watch the presentation, it was recorded here
Why most “evidence” doesn’t fit you
Think about every prevention headline you’ve read. “Reduces risk by X%.” Buried underneath is a study run on people nothing like you, on a Tuesday afternoon, in a room you’ll never sit in.
That’s not a knock on science. It’s a gap. On one side: the clinical trial, clean and controlled. On the other: you, with variable sleep, a wearable that sometimes doesn’t sync, supplements you remember most days, and months of messy, continuous, real data.
Almost nobody studies that second world. We built a platform that lives in it, so we could.
What we actually did
Over four months (October 2025 to February 2026), a group of APOE4 carriers added one thing to their lives: a transcranial photobiomodulation device, a helmet that sends gentle near-infrared light into the brain.
Then we measured everything around it. Six continuous streams per person: cognitive testing, sleep from Oura rings, blood biomarkers, HRV, daily check-ins, and device-usage logs. All captured inside one platform.
The cohort was high-risk on purpose: APOE4 carriers, 64% of them homozygous (two copies, the highest genetic risk there is). These are exactly the people most prevention research never reaches.
One honest note up front: different measures had different numbers of people, which is normal for real-world data. So judge each result against its own sample, not one headline number. We’ll tell you the sample every time.
How to read this honestly
Before the results, a promise: I’m going to give you the honest version, not the marketing version.
The study was small. That matters. A small study can only reliably catch big changes, so when a result just misses the cutoff, it’s usually real and simply waiting on a bigger sample to confirm. Keep that in mind as you read.
Three labels, that’s all you need:
Significant (p < .05) — strong, trust it.
Near-significant (p = .05–.11) — likely real, needs more people.
Stable — no meaningful change.
The headline: memory moved
Of the five thinking skills we tested, one stood out. And it’s the one that matters most.
Memory improved significantly. 20 of the 25 participants got better at it. The group’s memory score climbed from 62.96 to 70.32 (for the stats-minded: Z = 2.585, p = .010).
Of everything we could have hoped to move in APOE4 carriers, memory is the bullseye. It’s the domain most tied to early Alzheimer’s risk. Reasoning (p = .071) and perception (p = .106) leaned positive too, just short of the line a bigger study would likely cross. Attention and coordination held flat.
The wins weren’t random
Here’s the part that made me trust it.
Out of 22 individual skills we tested, four passed the strict bar on their own: naming, non-verbal memory, visual perception, and working memory. Five more were promising and near-significant.
Look at where they clustered: memory, naming, perception. The exact areas tied to early Alzheimer’s risk. Not scattered across random skills. That’s what a real effect looks like, not chance.
Most people improved
Zoom out to overall cognition and the picture holds: 60% of participants improved. The typical person’s gain was real, not statistical noise. The median change was +5.0, and the confidence interval ([2.0, 6.0]) excluded zero.
Was it uniform? No. Scores ranged from −19 to +21. Some people improved a lot, a few declined. That’s real life and real biology. But the center of gravity moved the right way.
Here’s the part most people miss
Stop for a second and notice how this study even exists.
No clinic. No research coordinator with a clipboard. The cognitive scores, the sleep data, the bloodwork, the daily check-ins, all of it came from carriers quietly tracking their own lives inside one platform. That platform is Phoenix.
That’s the whole idea. Phoenix is built by an APOE4/4 carrier, for APOE4 carriers, to turn scattered data into answers. 27 biomarkers tracked against APOE4-specific targets (not generic “normal”).
Monthly pods of carriers who keep each other consistent.
Experiments you run on yourself.
It’s the reason a study like this could happen at all, and why the next one will be bigger.
Hold that thought. Back to the data.
On session nights, sleep got better
We had one participant who tracked their sleep with an Oura ring, night after night, for months. The cleanest possible look at how one person responds.
On the nights they ran a session, every single sleep metric improved:
+32 minutes of total sleep
+10.5 minutes of REM
+7.8 minutes of deep sleep
13.7 fewer minutes to fall asleep
+6.6 points of sleep efficiency
+2.4 points of next-day readiness
All of it statistically solid (p < .01). The cleanest signals were REM and deep sleep.
One honest caveat: that’s one person, tracked deeply (n = 1). A strong pattern for them over time, not yet proof for everyone. Worth scaling, which is exactly what we plan to do.
And it built up the way a real effect does
This is the detail that’s hard to fake.
The benefits didn’t spike on day one like a placebo. They built. Daily readiness lifted by week 2 (+6.5 points). Total sleep time climbed by week 4 (+47 minutes), then strengthened from there.
Effects that grow over weeks are the pattern you’d expect from a real biological response.
A coherent story is forming
Insomnia symptoms eased too. On the standard index (lower = fewer symptoms), the typical participant dropped 2.5 points.
And here’s what gives it weight: better sleep architecture, fewer insomnia symptoms, and memory gains all leaned the same direction, in the same cohort. We even saw a hint that bigger insomnia drops tracked with bigger cognitive gains (a lead to test, not yet a finding, with only 8 people).
When independent measures agree, you start to believe the signal.
And people felt good, for four straight months
Across four members’ daily self-ratings (1,363 sessions in all) in the Phoenix app, sleep quality and mental sharpness trended up. Mood, energy, and overall wellbeing stayed high and steady the entire time.
That sounds modest. It isn’t. Mood and energy usually drift down over four months of effort and tracking fatigue. Holding them high and flat is itself a win.
(The one flag, in the name of honesty: self-reported stress crept up. Could be the device, could be life, could be the act of measuring yourself daily.)
The results that made us sit up
A couple of participants shared personal results outside the formal dataset. No control group, so we treat them as stories to test, not proof. But they’re striking.
One person’s IL-6, a key inflammation marker, fell more than 75% over about three months (6.1 → under 1.4 pg/mL). Their other inflammatory marker, already low, held steady.
Another improved by 5.7 points on the ADAS-Cog13, a standard Alzheimer’s cognitive assessment where lower is better (19 → 13.3).
Individual stories. But the kind that tell you exactly where to point the next, bigger study.
The scoreboard, honestly
Let me lay the whole thing out, nothing hidden:
Solid wins: memory improved (p = .010, 20/25 up); session-day sleep (all six metrics, n = 1); wellbeing held high for four months.
Promising signals: overall cognition (+5 median, near-significant); reasoning and perception trends; a sleep-to-cognition link worth testing at scale.
Watch closely: rising stress; the strongest sleep data being a single subject; wide individual variability.
Where it fell short
I won’t oversell this.
Here’s what held the study back: it was small and under-powered.
There was no placebo or sham group.
The strongest sleep data is from one person.
Follow-up was short.
We didn’t systematically capture life stress.
And repeat-testing can lift scores on its own.
So this isn’t absolute proof.
But it’s a real, encouraging signal, the kind that earns a tighter, larger study with a control arm.
That honesty is the point. You’ve been promised certainty by people selling you things your whole life. I’d rather show you the real picture and let you decide.
The device behind the results
The helmet in the study is the Neuronic Light, a clinical-grade transcranial photobiomodulation device. 1070nm near-infrared light, 256 LEDs, region-by-region control, a few minutes a day.
Want to try the exact device we studied?
We worked out a deal for this community: $100 off with code PHOENIX at checkout. It’s HSA/FSA eligible and ships with a 90-night home trial, so if it’s not for you, you send it back.
We used the Neuronic LIGHT device.
Want to be in the next study?
Here’s the thing the helmet can’t give you.
This study happened because thousands of APOE4 carriers decided to stop guessing and start measuring, together, in one place. 4,000+ carriers on Phoenix. 550+ actively running their own experiments. That’s the engine. That’s what turns a hunch into evidence.
When you join Phoenix, you’re not on the outside reading the next result. You’re in it. Your bloodwork, your sleep, your cognition become part of the dataset that proves what actually works for our genotype, while you get APOE4-specific tracking, a community of carriers who get it, and early access to studies like this one.
You carry APOE4. You can’t change that. But you can stop facing it alone, and start turning your own data into answers.
This is a movement, not a startup. Come build the evidence with us.
Dr. Kevin Tran,
Founder,
Phoenix (APOE4/4)
PS. Two ways forward, your choice. Want the device we studied? Code PHOENIX takes $100 off the LIGHT or the Neuradiant, with a 90-night trial so there’s nothing to lose. Want to help build the proof and get the tools for your own protocol? Join Phoenix. The carriers who do both tend to be the ones who, a year from now, actually know what’s working, instead of hoping.
Education, not medical advice. Findings are observational, single-cohort, partly single-subject (n = 1), with no control group, and are not proof of efficacy. Neuronic devices are not intended to diagnose, treat, cure, or prevent any disease. Consult your physician before starting any device or protocol. Phoenix may earn an affiliate commission on Neuronic purchases; it does not influence our recommendations or the study data.
