v1.4 · Jul 2026 · Retatrutide, CagriSema, VILPA · omega-3/rapamycin/NR demoted

A longevity operating system,
not a podcast transcript.

Every intervention ranked by the strength of its human evidence, its expected effect on healthspan, the cost, the inconvenience, and the organs it actually moves. Brain, heart, kidneys, liver, bone, muscle, immune, eyes — held to the same standard. Sauna, GLP-1s, resistance training, SGLT2, sleep, omega-3, hearing aids, rapamycin, peptides, nootropics — all on the same scoreboard.

Interventions
Real human outcomes
Biomarker-only
Mostly lore

If you do one thing for your brain, make it one of these.

The Lancet Dementia Commission (2024) estimates ~45% of dementia is preventable by addressing 14 modifiable risk factors. Below: every intervention in the OS with at least moderate human evidence for cognition or dementia, scored 0–3.

Scoring: 3 = hard outcome data (dementia incidence, cognitive decline slowed in RCT). 2 = moderate human data (cohort or biomarker + mechanism). Click any row for the full card.

What the newest evidence actually says.

eBioMedicine · Jun 2026 · New

Omega-3 (brain) demoted to LORE: DHA reached the brain and cognition didn't move.

The definitive CNS target-engagement RCT (Yassine et al., eBioMedicine, Jun 2026): n=365, 2 g/day DHA, 24 months. CSF DHA/AA ratio rose 0.19 (p<0.0001), independent of APOE ε4 — the supplement clearly reaches the brain. But no treatment differences in brain volumes or cognitive performance over two years. This is exactly the trial you'd design to settle the question, and it settles it. The cardiovascular/triglyceride claim is separate and unchanged; the brain claim is now LORE.

Lilly · May & Jul 2026 · New

Retatrutide readouts: −28.3% weight over 80 weeks, with a directional MACE signal.

Lilly's triple agonist (GLP-1/GIP/glucagon) reported all three TRIUMPH Phase 3 trials in 2026. TRIUMPH-1 (n=2,339, 80 weeks): weight change −19.0% / −25.9% / −28.3% (4/9/12 mg) vs −2.2% placebo; 45% of the 12-mg arm hit ≥30% loss. TRIUMPH-3 five-component MACE HR 0.82 (0.55–1.22) — directional, underpowered. Board adds retatrutide as REAL for metabolic, MIXED for hard longevity endpoints until a proper outcomes trial reports. Same class caveats: pair with resistance training and ≥1.2–1.6 g/kg protein.

Aging · Apr 2025 · New

PEARL: rapamycin missed its healthy-aging primary endpoint at p=0.942.

PEARL — the only dedicated healthy-aging rapamycin RCT — published in Aging (Albany NY): 48 weeks, weekly 5 mg or 10 mg vs placebo. Primary endpoint (visceral adiposity) not met, p=0.942. Secondary: lean tissue mass ↑ (ηp²=0.202, p=0.013), pain ↓ in women on 10 mg. Interesting biology; no hard-outcome case for healthy adults. Verdict moves from MIXED to LORE for the healthspan indication.

Three 2025 RCTs · New

Nicotinamide riboside demoted to LORE: NAD⁺ doubled, everything else was null.

Three 2025 RCTs on the same molecule tell the same story: NAD⁺ raising is confirmed, downstream endpoints are not. aMCI phase-II (Alz Dement, Dec 2025, n=42, 12 wk, 1 g/d) — blood NAD⁺ 23→48 µM (p<0.001); cognition not improved. NR + exercise for hypertension (GeroScience, Dec 2025, n=54): daytime systolic BP +5.2 vs −2.7 mmHg — NR was worse. SCD/MCI crossover (Alz Dement TRCI, Jan 2025) — no RBANS change; only pTau-217 moved. Textbook "don't fool yourself."

BJSM · Jun 2026 · New

Resistance training joins evidence-5: 147k people, 30 years, HR 0.55 combined.

The largest pooled dose-response analysis to date (BJSM, Jun 2026): n=147,374, up to 30-year follow-up, 35,798 deaths. 90–119 min/week of resistance training → all-cause mortality HR 0.87 (0.81–0.95); cardiovascular mortality HR 0.81; neurological-disease mortality HR 0.73. Combined with aerobic activity: HR 0.55 (0.50–0.60). The sweet spot is 90–120 min/week and the biggest gain is stacking with aerobic work. Now the highest-evidence non-drug lever on the board — and even more essential now that GLP-1s are pulling weight down at the cost of lean mass.

IJBNPA · Jan 2026 · New

VILPA: 1–2 minute exercise "snacks" 5–8×/day map to HR 0.46 for mortality.

Accelerometry cohort (IJBNPA, Jan 2026, n=3,293 US adults, 6.7-year follow-up, 290 deaths): 5.3 vigorous-intermittent-lifestyle-activity bouts/day → all-cause mortality HR 0.56 (0.39–0.82); ~8 bouts/day → HR 0.46 (0.28–0.77). Backed by an 11-RCT meta on VO₂max and 60-second sit-to-stand. Confounding by baseline capacity is plausible — but for people who won't do structured training, this is the highest-yield entry we've ever added. New on the board in v1.4.

Cell Rep Med + Neurology · 2026 · New

Hearing aids upgrade: three converging datasets, RR 0.67 for 7-year dementia.

Pooled 7-cohort analysis (Cell Rep Med, May 2026, n=61,089, 6.5y, 8,911 events): hearing-aid use HR 0.91; effective hearing improvement HR 0.86; poor correction HR 0.98 (i.e., wearing them matters). Target-trial emulation (Neurology, Feb 2026, 7-year): dementia risk 5.0% vs 7.5% (RR 0.67, 0.37–0.97), inverse dose-response by use frequency. ACHIEVE secondary analysis: 61.6% slower decline in the highest-risk quartile. Ownership without use doesn't count.

Clin Nutr + BMJ · 2026 · New

Vitamin D is the archetype of a conditional REAL: works only if you're deficient.

Target-trial emulation of RCTs (Clin Nutr, Mar 2026, n=237,502 and 185,809, 5.3–5.7y): overall all-cause mortality NULL (HR 0.97 and 1.02); in insufficiency HR 0.85 and 0.81; in deficiency HR 0.79 and 0.75. Fractures (BMJ, May 2026, 69 trials / 153,902 participants): vitamin D alone RR 1.00, high certainty. Measure 25(OH)D, correct if low, do not megadose a sufficient level. B-vitamins behave the same way — real but small, and only in the elevated-homocysteine framing.

CJASN · Feb 2026 · Updated

Semaglutide splits by endpoint: REAL for cardiometabolic + kidney mortality, LORE for dementia.

FLOW severity analysis (n=3,533 T2D+CKD, 3.4y): all-cause death HR 0.80 (0.67–0.95); UACR ≥2000 mg/g subgroup HR 0.47 (0.31–0.70). SELECT hospitalization ratio 0.90 (p<.001). But EVOKE / evoke+: oral semaglutide 14 mg did not slow cognitive or functional decline in early Alzheimer's. File the AD claim under "does not work for."

JAMA · Aug 2025 · Anchor

US POINTER: structured lifestyle beats self-guided — statistically robust, clinically small.

US POINTER (n=2,111, 2 years): the structured MIND-style diet + supervised exercise + cognitive training + cardiometabolic monitoring bundle improved global cognition 0.243 vs 0.213 SD/year self-guided — between-group Δ 0.029 SD/year (95% CI 0.008–0.050, p=0.008). Both arms improved. The active ingredient is the bundle. MIND diet upgraded to evidence-4; effect size honestly labeled Small.

BMJ + Lancet HL · 2026 · New

Menopausal HRT: small mortality signal, dementia claim now formally null.

Register-based cohort (BMJ, Feb 2026, n=876,805, 14.3y): adjusted death HR 0.96 (0.93–0.98); 3–4.9 y of use HR 0.90 (0.84–0.95). Dementia (Lancet Healthy Longev, Dec 2025, n=1,016,055 across 10 studies): no association between MHT and MCI/dementia — subgroup analyses by timing, duration, formulation also null. REAL for symptom control and possibly small mortality benefit; LORE for the "HRT prevents dementia" narrative.

Centenarian resilience · 2026

Cognitively healthy centenarians are genetically protected against Alzheimer's — not just lucky.

A five-cohort analysis shows centenarians across the US, Europe, and Asia carry a higher Alzheimer's polygenic protective score than controls — beyond APOE. Protection increases with age: supercentenarians (≥110) carry the most protective alleles. Complementary work on centenarian microglia finds they overexpress neprilysin, clearing amyloid 1.5–2× faster than young adult microglia. Implication: reaching 100 cognitively intact is a resilience phenotype, not an absence-of-risk one.

The operating system.

Ranked by a composite of evidence strength, effect size, and biomarker coverage — penalized by cost and inconvenience. Filter to see where you want to spend your time.

Verdict
Category
Evidence from
# Intervention Verdict Evidence Effect Cost / mo Effort Biomarkers Composite

Every intervention × every organ.

How many strong-evidence interventions actually target each organ. Cells score 0–3: blank = none, 1 = plausible, 2 = moderate human data, 3 = strong human outcomes.

What actually has human outcome data — and what doesn't.

The hierarchy: REAL interventions have hard human outcomes (mortality, disease, function). MIXED have only biomarker or short-term RCTs. LORE is mostly preclinical, anecdotal, or podcast culture. An intervention can be worth doing at any level — but you should know which shelf you're buying from.

Click any intervention for the full card.

If you had to pick a starting stack,
start where the outcomes are.

Tier 1 · Do these first
  • Resistance training — 90–120 min/week, stacked with aerobic work (BJSM 2026: combined HR 0.55)
  • Sleep ~7.2 h self-reported / ~7.7 h in bed, consistent schedule; screen for OSA if loud snorer
  • VO₂max work — 150+ min/week moderate or 75 min vigorous
  • VILPA / exercise snacks — 5–8 bouts/day of 1–2 min vigorous incidental activity (new in v1.4)
  • Creatine 3–5 g/day for muscle/strength (kidney check if CKD)
  • Hearing aids if hearing is impaired — wear them consistently (RR 0.67 for 7-year dementia)
  • Eggs ≥5/week — 27% lower AD risk in 40,000-person Loma Linda cohort

Each has a mortality or dementia signal in humans. Nothing else on this site has a stronger case.

Tier 2 · Add if they fit your life
  • Structured multi-domain lifestyle bundle (MIND-style diet + supervised exercise + cognitive training + cardiometabolic monitoring) — US POINTER 2025 showed the bundle beats any single component
  • Sauna — 4+ sessions/week, ≥19 min (KIHD signal holds; 2025 mechanistic RCTs null — downgraded to MIXED in v1.4)
  • Omega-3 from fatty fish 2×/week — supplement only for triglycerides/CV; brain claim is now LORE (eBioMedicine 2026)
  • Time-restricted eating — 8–10h window as an adherence scaffold, not a metabolic drug (isocaloric Δ −1.5 kg)
  • Vitamin D3 titrated to a sufficient 25(OH)D — only if measured deficient (Clin Nutr 2026: works only in the deficit)
  • Menopausal HRT — for symptom control and possibly small mortality benefit; not for dementia prevention

Solid biomarker or symptom evidence; hard-outcome case is smaller than the podcast conversation suggests.

Tier 3 · Consider with a physician
  • GLP-1s (semaglutide, tirzepatide) — when BMI/metabolic risk warrants; sema has the clearest kidney-mortality case (FLOW HR 0.80)
  • Retatrutide / CagriSema — next-generation weight/metabolic pharmacology (−28% weight over 80 wk); no hard-outcome trial yet
  • SGLT2 inhibitors — first-line for CKD ± diabetes, heart failure protection
  • Bempedoic acid — for statin intolerance; CLEAR Outcomes MACE HR 0.87 (all-cause death was flat)
  • Colchicine 0.5 mg — for established CAD; 2025 meta-analyses split on CV-mortality benefit
  • Resmetirom — for biopsy-confirmed MASH with fibrosis; hard-outcome readout at 54 months not yet
  • B-vitamins — only if homocysteine elevated (high-certainty g=0.11 — real but very small)
  • Metformin — for T2D/prediabetes; healthy-longevity indication is LORE (TAME still pre-launch)
  • AREDS2 — only for intermediate AMD
  • GHK-Cu cream — topical only, skin outcomes

GLP-1s and their triple-agonist successors have the strongest hard-outcome data of any Rx here. Mouse-to-human longevity translation for the rest is still an open bet.

Tier 4 · Treat as entertainment, not a plan
  • NR, NMN — NAD⁺ rises; three 2025 RCTs on NR all null for cognition, BP, function (v1.4 demotion)
  • Rapamycin for healthy aging — PEARL primary endpoint missed at p=0.942 (v1.4 demotion)
  • Omega-3 supplements for brain — target-engagement RCT: DHA reached the brain, cognition didn't move
  • Epitalon, Thymosin α-1 for longevity — no healthspan trials; oncology-adjunct evidence only
  • Cold plunge beyond short cold showers
  • BPC-157, CJC/Ipamorelin — newest BPC-157 human paper is n=2; FDA-restricted
  • Fisetin/senolytics for humans — 74-person knee-OA RCT fully null; taurine, urolithin A, klotho — same story
  • Lion's Mane, Bacopa, most nootropic stacks

May feel good. Probably won't move your 10-year mortality curve. Some have real downside risk.

The primary sources this is built on.