Phoenix research is real-world and observational. We follow real APOE4 carriers over time and compare each member with their own earlier self. No placebo group, no randomisation: members chose their own interventions, and we show you everything else they changed alongside them.
For carriers, this is deliberately more pragmatic than academic research. The goal is not to force you to run one intervention in isolation. It is to record every change well enough to identify which interventions are helping and which are not. Phoenix currently shows the observations and everything else that changed alongside them. Phoenix is building pooled analysis that will combine timing, dose, adherence, repeated personal measures, and de-identified community patterns. When the data cannot separate two changes, Phoenix will say that plainly.
See the current Release 001 publication status in APOE4: The Beat the Odds Study. The searchable Research Navigator is open: every reviewed result is searchable by goal, marker, or intervention.
Members log bloodwork, supplements and medications with start dates, daily adherence, daily check-ins, wearable sleep and heart-rate variability, and cognitive-game sessions. We link those streams together and look for what changed, when, and by how much.
The core comparison is within-member: a member against their own earlier self, or their own other days. That is deliberate. It removes every difference between people, age, genotype, sex, income, baseline health, diet, doctor, because both sides of the comparison are the same person. It is the single strongest thing about this design.
Three kinds of finding get published
One member, one timeline, one interesting change.
The same direction repeating across several eligible members.
Things we expected to find and did not. We publish these too.
Grouped under design, measurement, statistics, and safety and scope. Each one is a real way this research can mislead you, and each is the reason we read every number below beside its caveat rather than alone.
There is no placebo group.
Nobody in Phoenix was given a dummy pill. When a member's number improves, some of that improvement can be expectation, attention, and the simple act of paying attention to your health. Placebo effects on subjective ratings like energy and mood are large and well documented. They are smaller, but not zero, on objective measures like ApoB and wearable sleep duration.
Nothing is randomised.
Members chose what to take and when to start. That choice is not random, and the reason for the choice can be the real cause of the result.
Confounding by indication.
People often start an intervention because a number is already moving. Someone starts berberine because their lipids are climbing. Someone starts a sleep protocol during a bad month. Both make the intervention look worse than it is. The reverse also happens: someone starts a supplement during a motivated, everything-is-going-well stretch, which makes it look better than it is.
Co-intervention, and it is severe here.
Phoenix members stack. Members with repeat bloodwork log an average of 12 concurrent interventions; members with no bloodwork log 0.4. So “has repeat bloodwork” and “is a heavy stacker” describe nearly the same people, and any single-intervention signal could be the rest of the stack talking. For every individual case we publish, we state how many other interventions that member started nearby. When that number is zero, the case is much stronger. We say which.
Self-selection.
Phoenix members are not a random sample of APOE4 carriers. They are people who found out their genotype, sought out a community, and paid to join it. 59% carry two copies of APOE4, against about 2.7% in the general population, about a 22-fold enrichment. That makes Phoenix an extraordinary place to study APOE4 and a poor place to estimate what is typical.
Survivorship.
The members with the longest data histories are the ones who stayed. People for whom nothing worked are more likely to have stopped logging, which quietly tilts every long-window result positive.
Daily check-ins are subjective.
Sleep quality, energy, mood, mental sharpness, calm and wellbeing are member-entered ratings on a 1–10 scale. They are real and they matter, but they are a person's impression of their day, not an instrument reading.
Some fields do not mean what their name says.
In the check-in data, mood and overall wellbeing are the same underlying field, so we never report them as two independent confirmations. The field named stress behaves as a calm score where higher is more relaxed. We publish these quirks rather than quietly working around them.
Lab noise can be bigger than the effect.
Every blood test has test-retest variability. Where a change sits inside the assay's own expected variation, Phoenix leaves it out of the published result set.
Units and naming are a minefield, and it has bitten us.
The same biomarker arrives from different labs under different names and in different units. ApoB and apolipoprotein_b are the same thing; LDL cholesterol, LDL particle number and LDL size are three different physical quantities that a careless match will average together into nonsense. Genotype arrives spelled four different ways. One naive match once discarded 866 of 1,277 members without any error appearing. We now canonicalise names, convert units explicitly, and print what we folded together.
Timing and lag are uncertain.
We know when a member logged a start. We do not always know when they actually started, whether they took it consistently, or at what dose. Dose is recorded on about 57% of entries and time of day is not captured at all. A follow-up blood test 77 days later is a real measurement at a real time, but the exposure between those two points is partly inferred.
Coverage is incomplete.
Not every member wears a device, retests on schedule, or logs every day. Missing data is rarely missing at random: people log more when things are going well.
Multiple comparisons.
When many intervention-and-outcome combinations are tested, some will look interesting by chance. Where Phoenix reports a statistically adjusted result, it names the adjustment. Individual cases are presented as individual cases.
We slice the group to find the signal, and we tell you which slice.
When we report a pattern, we have often looked at a subgroup: one genotype, one sex, an age band, the members whose starting number was worst. Choosing the cut after seeing the data makes a finding easier to produce by chance, and we are not going to pretend otherwise. Our answer is not to stop doing it, because a signal in a subgroup is exactly what a carrier wants to know about. Our answer is that the cut is always named in the sentence. If a number describes twelve women over 55, it says so. If it describes everyone, it says that too. Read the cut before you read the number.
Small numbers.
Many of these cohorts are single or double digits. An n=4 signal where all four move the same way is genuinely interesting and genuinely fragile. Both things are true.
Regression to the mean.
People often start something when a number is at its worst. The next reading tends to be closer to their average regardless of what they did. This alone can manufacture an impressive-looking improvement.
Underpowered is not the same as ineffective.
A small or incomplete cohort may be unable to detect a real effect. A null result means Phoenix did not see an effect under that exact analysis; it is not evidence that the intervention is ineffective.
There is no adverse-event surveillance system.
Phoenix does not run structured safety monitoring. We cannot claim that no serious adverse events occurred among members, only that none were reported to us through the app.
Device and partner studies carry two extra limits.
Wearables are often not connected before a member's first use, so there is no clean group baseline to measure against. The comparison has to be use-nights against that member's own non-use nights instead. And when we report a device study, we report the discomfort too: in the ZenoWell taVNS study, alongside the positive sleep-onset reports, five members used pain language, two reported ear or skin irritation, two reported headache, and three used stop language. Positive experiences, null experiences and adverse experiences all get published.
This is not medical advice.
Phoenix does not diagnose, treat, cure or prevent any disease, including Alzheimer's. Medication decisions stay with your clinician. Phoenix does not check interactions with anything you already take. Nothing here is a recommendation to start, stop or change a treatment.
Every number in APOE4: The Beat the Odds Study follows these same rules.
The frozen cohort.
Every Release 001 community-wide statistic uses the same 476 Phoenix members: membership active on 25 August 2026, joined on or before 23 August 2026. A chart naming a smaller n draws from within that same cohort.
The threshold rule.
A marker counts as improving or worse when it moves 5% or more in that direction between a member's first and latest recorded result; anything smaller counts as steady. HbA1c moves in smaller steps, so it uses its own line instead: a 0.1 percentage point move or more.
The data-through boundary.
Bloodwork and intervention statistics run through 23 August 2026. Wearable and check-in analyses run through 9 August 2026, an earlier boundary because that data arrives on its own schedule.
The pragmatic stack rule.
For an individual carrier, the useful question is whether the complete stack improved the target result. Several components can contribute and can work together, so Phoenix shows the full recorded stack and highlights the interventions most likely to have contributed rather than crediting exactly one.
Why Phoenix leaves some results out
None of the above is unique to Phoenix. It describes the majority of the nutrition, supplement and lifestyle research that reaches the public, the studies behind the headlines, the podcast claims and the supplement labels. Most of it is observational. Most of it has no placebo arm. Most of it relies on people remembering what they ate.
The record of what happens when those observational findings finally get tested properly is sobering:
| Observational finding | What the randomised trial found |
|---|---|
| High beta-carotene intake tracked with LESS lung cancer | ATBC (1994), 29,000+ male smokers: 18% MORE lung cancer and 8% higher overall mortality in the beta-carotene arm. CARET (1996), 18,314 participants: 28% more lung cancer, 17% higher mortality. CARET was stopped early and participants were told to stop taking their vitamins.ATBC, NEJM 1994CARET, NEJM 1996 |
| Vitamin E tracked with LESS heart disease | HOPE (2000), 9,500+ high-risk patients over four years: no cardiovascular benefit. GISSI-Prevenzione, 11,000 heart-attack survivors over three-plus years: no preventive effect.HOPE, NEJM 2000GISSI-Prevenzione, Lancet 1999 |
| Hormone therapy tracked with LESS coronary heart disease across large observational cohorts | The Women's Health Initiative (2002), 16,608 healthy postmenopausal women, stopped early at 5.2 years: more coronary events and more invasive breast cancer in the treated arm. The field then spent two decades reconciling the two, and the timing of initiation turned out to matter enormously.WHI, JAMA 2002 |
The lesson is not that observational research is worthless. It is that observational research is how you find the question, and a controlled trial is how you answer it. Phoenix is deliberately in the first business, and we are explicit about which business we are in.
What we will not do is what most of the sources our members already read do do: present an observational pattern with the confidence of a trial result, and leave everything else that changed out of the picture.
These are the actual defences in the pipeline, all of which have caught real errors.
Within-member comparison, always. Every member is their own control group. It is not a placebo arm, but it eliminates every between-person difference in one move.
The support-floor test. A real within-member effect gets stronger as you demand more days of evidence per member. Noise gets weaker. A stretching-improves-energy result did not meet this test’s review standard and was not published as a finding. Phoenix publishes results that meet every review criterion.
Placebo tags as a control arm. We run mechanistically inert tags through the identical pipeline. When a vitamin with no plausible acute mechanism out-scored physical exercise, we knew the design was generating the signal, not the intervention. That result was not published.
Positive controls. We run a known drug effect through the pipeline to prove the method can detect something real before we trust it on something new.
We check our own analysis. Every edition is analysed independently more than once and reconciled line by line before anything is published. Only reviewed results appear here.
We publish the nulls. Release 001 found no supplement with community-level efficacy. Promising individual results remain individual results until comparable protocols repeat across more members.
An individual case is one person. It tells you something interesting happened to somebody real, with dates and numbers attached. It does not tell you it will happen to you.
Read the co-start count. Zero means the member changed one thing. Eleven means you are looking at a whole protocol, and the result belongs to the protocol.
Read the elapsed time. A 40% ApoB drop over 77 days and over 400 days are different stories.
Direction repeating across members is worth more than size in any one member. Four of four moving the same way beats one dramatic case.
When we say “we could not see it”, that is not “it does not work”. See limit 16.
Every signal we publish comes with the test that would make it stronger. In general that means complete start and stop dates, recorded dose and adherence for every intervention, a pre-specified retest date, repeated personal measurements, and where practical staggered or alternating timing. Phoenix is building pooled analysis that combines those records with de-identified community patterns to identify what is helping, what is not, and what remains unresolved.
Phoenix reads broad findings as evidence to weigh, not instructions to copy. Study design, population, effect size, and uncertainty determine how much confidence any finding deserves.
Solomon et al. (2018), JAMA Neurology
The FINGER subgroup analysis found that APOE4 carriers could benefit from a combined program of nutrition, exercise, cognitive training, and vascular monitoring. The carrier-versus-non-carrier interaction was not statistically significant, so Phoenix treats this as a promising signal, not proof of a larger effect.
Kaufman et al. (2021), Journal of Cerebral Blood Flow & Metabolism
A 52-week randomized study linked aerobic exercise with changes in hippocampal blood flow in hypertensive APOE4 carriers. The subgroup was small, making replication and individual measurement important.
Fortea et al. (2024), Nature Medicine
A large multicohort analysis described a more predictable biomarker timeline among APOE4 homozygotes. It strengthens the case for genotype-aware research, but it does not determine any one person’s future.
Physicians, Doctor of Pharmacy specialists, and APOE4 researchers challenge interpretation, safety, and overreach. Phoenix educational material does not replace advice from a qualified clinician who knows your health history.
These grouped patterns come first because repetition across members is more informative than one person's result. Where a real alternative explanation exists, Phoenix names it beside the number.
Statin member case
APOE3/4 · 62 · femaleMember #5AA27F lowered ApoB from 98 to 59 mg/dL in 77 days; generic and named statin use counts are not summed.
Curated statin-family cases with a dated start and repeat ApoB result.
Sleep-timing signal
TWO SEPARATE ANALYSESTwo analyses found earlier nights about 30 minutes longer: 56 of 68 eligible members under one rule and 59 of 66 eligible members under a stricter rule. No blended response rate is published.
Members eligible under each named within-member median-bedtime analysis; the two cohorts remain separate.
Sleep-and-day signal
n=32 · PHOENIX MEMBERSEnergy, mood, sharpness, and calm were higher after longer nights
Within-member comparisons linked objective sleep with the next daily check-in, reducing some between-person differences.
Routine-linked signal
n=4 · PHOENIX MEMBERS+0.57 energy and +0.60 wellbeing points
Each member was compared with their own explicitly missed days. All four member-level differences were positive for both energy and wellbeing.
Strongest caveat: Magnesium days also included 6.17 more logged interventions on average. This is more likely a whole-routine signal than an isolated magnesium effect.
These are within-member taken-versus-explicitly-missed-day differences. They are not isolated supplement effects. On focal taken days, members usually logged several other interventions too. Phoenix treats them as whole-routine leads for prospective testing.
Energy +0.28 · wellbeing +0.50 · sleep +0.64
On taken days: +8.07 other interventions on average.
Wellbeing +0.47 · sleep +0.42
On taken days: +6.82 other interventions on average.
Wellbeing +0.68 · sleep +0.52
On taken days: +9.33 other interventions on average.
Energy +0.37 · sleep +0.76
On taken days: +1.55 other interventions on average.
Energy +0.31 · wellbeing +0.81
On taken days: +13.71 other interventions on average.
Energy +0.55 · wellbeing +0.78 · sleep +0.67
On taken days: +11.40 other interventions on average.
Randomized, placebo-controlled trials are essential for estimating whether an intervention causes an effect under controlled conditions. They take time and resources, and they deliberately study defined populations.
Phoenix asks a complementary question: what patterns appear in real APOE4 carriers living ordinary, disrupted lives? Our observations are faster and more personal, but they are also noisier and cannot replace controlled trials. The strongest result is a signal worth testing better, not a verdict.
In biomarker-confirmed Alzheimer's clinical-trial cohorts, 65% to 70% of participants carry APOE4. Yet even a good trial rarely answers the question each carrier faces at home: what works for me, in real life?
You can take ten supplements, change your diet, add a medication, and still not know what helped. Multiple changes are not the problem. Unrecorded changes are. Phoenix records the timing, adherence, measurements, and context needed to interpret the full stack.
Phoenix is building pooled analysis to help identify what is helping, what is not, and what the data cannot separate yet.
More certainty for youGenetics · baseline · context · adherenceWhat should I test?
Clinical trials, research papers, Phoenix Member results, and your data become one candidate insight with a confidence level.
Phoenix AIAPOE4 evidence, personalized with your data.The evidence fits. Your personal response is still unknown.
How do I test it?
Name the outcome. Log starts, stops, dose, adherence, and context. Measure before and after.
Did it work for me?
If the result moves as expected, confidence may rise. If not, you know what to reconsider with your clinician.
You now have personal evidence to keep, confirm, or discuss with your clinician.
Phoenix lowers confidence in this candidate and helps you choose the next test.
What becomes possible next?
The goal is to combine your repeated measurements with de-identified patterns from other members, while accounting for timing, baseline, environment, adherence, and real-life context.
The more well-measured interventions members record, the more evidence Phoenix can use beyond population averages.
Phoenix AI already learns from published APOE4 research, clinical-trial evidence, Phoenix Member data, and your own results.
What appears to work for people like you may be worth testing for you. What appears to work for you may help someone similar ask a better question.
Help yourself find what works for you, and help APOE4 carriers around the world advance real-world research. Donations help Phoenix run more prospective n=1 studies, strengthen this shared learning loop, and turn one member's result into a better question for the next.
To support the work, email kevin@thephoenix.community.
A Phoenix check-in is a short, member-entered snapshot of daily life: sleep quality, energy, mood, mental sharpness, calm or stress, and overall wellbeing. It is subjective, so Phoenix interprets it beside objective records rather than alone.
Data were analyzed independently three times, adjudicated, and rerun against the underlying records.
A lab, wearable measure, cognitive task, or check-in before the change.
A supplement, medication, behavior, or protocol begins.
Was it taken? How often? What else changed at the same time?
What changed next, when, and by how much?
Does another eligible Phoenix Member move in the same direction?
The signals above are the starting point. The next step is a pre-planned experiment that changes one thing, measures adherence, retests the outcome, and updates what Phoenix knows about you.
Which intervention moved your ApoB, and did the combination do more than either alone?
Medication choices and any sequential add-on design remain clinician-guided.
Research opportunity
44 Phoenix Members have recent ApoB above 70 mg/dL and no lipid medication logged.
Measure: ApoB, LDL-C, non-HDL-C, adherence, diet, weight, and every dated intervention change.
Does it change your cognitive performance, your energy, both, or neither?
Repeated baselines help separate an intervention signal from practice effects and better days.
Measurement base
Phoenix already has 167 completed cognitive sessions, but no closed intervention sequence.
Measure: Repeated cognitive tasks, daily energy, sleep, dose, adherence, and a pre-set 12-week endpoint.
Does it improve triglycerides and cognition enough to justify the time for you?
A prospective design can test the intervention itself instead of comparing naturally active days.
Signal to challenge
Across 15 members, higher-activity days averaged 6,882 more steps. The next-day analysis was null.
Measure: Assigned sessions, heart-rate zone, adherence, triglycerides, sleep, and repeated cognition.
Release 001, APOE4: The Beat the Odds Study, is out: 43 pages built from the reviewed result set. Enter your email on the results page and it is yours instantly.