Hypothesis: Senescent immune cells export oxidized mitochondrial DNA via exosomes to drive systemic aging through cGAS-STING activation

The immune system accelerates aging not only by secreting inflammatory cytokines but also by transferring damaged mitochondrial components to distant tissues. Recent work shows that senescent immune cells from aged mice induce widespread senescence when transplanted into young hosts[https://doi.org/10.1038/s41586-021-03547-7] and that hematopoietic DNA damage alone can prematurely age solid organs[https://doi.org/10.1038/s41586-021-03547-7]. Here we propose that the key vehicle is oxidized mitochondrial DNA (ox-mtDNA) packaged into exosomes released by senescent CD28⁻ T cells, S100A8⁺ neutrophils, and aged monocytes. Ox-mtDNA acts as a damage‑associated molecular pattern that activates the cytosolic cGAS‑STING pathway in parenchymal cells, leading to chronic type‑I interferon signaling, NAD⁺ depletion, and reinforcement of the senescence-associated secretory phenotype (SASP). This creates a feed‑forward loop: immune‑cell‑derived ox‑mtDNA amplifies tissue senescence, which in turn releases more DAMPs that further activate immune cells.

Testable predictions

1. Plasma levels of exosome‑associated ox‑mtDNA will correlate with biological age across mouse strains and human cohorts, and will be elevated in conditions of accelerated immunosenescence (e.g., Vav‑iCre Ercc1 mice).
2. Pharmacologic inhibition of exosome release (e.g., 'GW4869') or neutralization of extracellular mtDNA with anti‑DNA antibodies in aged mice will reduce tissu

Hypothesis

Pain sensitivity measured at the circadian nadir (03:00–04:00) reflects NAD+-dependent SIRT1 activity, which in turn modulates epigenetic age acceleration. It's reasonable to expect that individuals with robust circadian NAD+ rhythms will show higher pain thresholds at night and slower GrimAge2 advancement, whereas flattened NAD+ rhythms predict low nocturnal pain tolerance and accelerated epigenetic aging.

Mechanistic Rationale

• Circadian NAD+ oscillations drive SIRT1 deacetylation of histones and clock genes, linking metabolic state to chromatin modifications that underlie epigenetic clocks (https://pmc.ncbi.nlm.nih.gov/articles/PMC3852564/).
• SIRT1 activity suppresses NF‑κB signaling and reduces inflammatory cytokine production, thereby lowering nociceptor sensitization (https://www.tandfonline.com/doi/full/10.2147/JPR.S521606).
• Pain perception follows a ~24‑hour rhythm with lowest tolerance at 03:00–04:00, driven by the suprachiasmatic nucleus (https://medicalxpress.com/news/2022-09-link-pain-sensitivity-circadian-rhythm.html).
• Aging dampens NAD+ amplitude and blunts SIRT1 activity, which would simultaneously raise nocturnal pain sensitivity and accelerate DNAmGrimAge2 (https://pmc.ncbi.nlm.nih.gov/articles/PMC6710702/).

Thus, nocturnal pain threshold can serve as a peripheral read‑out of NAD+/SI

The observed drop in the Firmicutes/Bacteroidetes ratio after age 70 represents an adaptive host response that attempts to rebuild the anaerobic lumen by boosting colonocyte butyrate oxidation. This hypothesis predicts that activation of peroxisome proliferator‑activated receptor alpha (PPARα) in aged colonic epithelial cells increases fatty acid β‑oxidation, thereby consuming luminal oxygen and recreating the low‑oxygen niche favored by obligate anaerobic butyrate producers. Restoration of this oxygen gradient should suppress facultative anaerobes and Gammaproteobacteria, raise the F/B ratio toward youthful levels, and improve barrier function.

Butyrate normally supplies 70‑80% of colonocyte ATP through mitochondrial β‑oxidation, a process that directly consumes oxygen and maintains a serosal‑to‑luminal O2 gradient that keeps the lumen anaerobic [https://pmc.ncbi.nlm.nih.gov/articles/PMC8002420/]. Age‑related loss of butyrate‑producing microbes reduces this oxidative flux, leading to colonocyte energy deficit, AMPK activation, and a breakdown of the oxygen barrier that permits oxygen‑tolerant pathobionts to expand [https://www.frontiersin.org/journals/aging/articles/10.3389/fragi.2025.1452917/full]. PPARα agonists are known to upregulate genes involved in fatty acid transport and β‑oxidation, enhancing mitochondrial respiration without necessarily increasing mitochondrial mass. By pharmacologically stimulating PPARα in aged colonocytes, we can compensate for the microbial deficit in butyrate supply, increase oxygen consumption, and thereby force the lumen back into an an

Hypothesis

Berberine’s longevity benefits stem from AMPK‑driven selective mitophagy that purges mitochondria carrying high levels of pathogenic mtDNA heteroplasmy, thereby lowering the mutational load below a tissue‑specific threshold and breaking the vicious cycle of deletion‑mutant amplification.

Mechanistic Rationale

1. AMPK activation by berberine phosphorylates the mitophagy receptors FUNDC1 and BNIP3, increasing their affinity for LC3 on depolarized mitochondria.
2. Depolarized mitochondria are enriched for mtDNA lesions and high heteroplasmy because damaged genomes impair oxidative phosphorylation and membrane potential.
3. Selective removal of these organelles reduces the fraction of mutant mtDNA copies without triggering compensatory biogenesis, as PGC‑1α activation is coupled to a quality‑control checkpoint that senses reduced ROS and restored NAD⁺/NADH ratio.
4. Consequently, the net mtDNA copy number may stay stable or even decline in diseased cells, while healthy repopulation occurs from the remaining low‑mutagen load.

Testable Predictions

• Prediction 1: In mtDNA mutator (PolG⁻/⁻) mice, chronic berberine treatment will decrease heteroplasmy at common deletion sites (e.g., mtDNA4977) in skeletal muscle and brain by ≥30 % compared with vehicle, measured by droplet digital PCR.
• Prediction 2: The heteroplasmy reduction will correlate with improved respiratory control ratio (RCR) and decreased ROS emission, whereas total mtDNA copy number remains unchanged or slightly reduced.
• Prediction 3: Muscle‑specific AMPKα1/α2 double knockout will a

Hypothesis

Core claim: In aged tissues, a subset of senescent myeloid‑derived suppressor cells (MDSCs) functions as a short‑lived "emergency reserve" that, via NLRP3‑driven IL‑1β and IL‑6 secretion, primes hematopoietic stem and progenitor cells (HSPCs) to undergo emergency myelopoiesis. This transient boost enhances innate immune responsiveness to acute challenges (e.g., infection, tissue injury) but becomes maladaptive when the senescent MDSC pool persists, fueling chronic immunosuppression.

Mechanistic rationale

• Senescent MDSCs exhibit high p21 and secrete IL‑1β through active NLRP3 inflammasome [NLRP3 inflammasome activation drives senescence through IL-1β secretion].
• Their SASP also contains IL‑6, which acts on MDSCs themselves via MAPK to sustain their immunosuppressive phenotype [IL-6 promotes the proliferation and immunosuppressive function of myeloid-derived suppressor cells via MAPK signaling]
• IL‑6 and IL‑1β are known to stimulate HSPC proliferation and myeloid bias through STAT3 and NF‑κB pathways, a process documented in emergency myelopoiesis [Emergency myelopoiesis is driven by IL‑6 and IL‑1β signaling in HSPCs]
• Thus, senescent MDSCs create a paracrine loop that temporarily expands granulocyte‑monocyte progenitors, increasing neutrophil and inflammatory monocyte output.

Predictions and falsifiability

1. **Transient NLRP3 inhibition in aged mice after LPS challenge will reduce emergency gra

Hypothesis

We propose that mitochondrial ROS accumulation in aged c‑Kit+ cardiac progenitors activates the NLRP3 inflammasome, driving a senescent state marked by p16INK4A upregulation and a pro‑inflammatory SASP. This axis creates a feed‑forward loop that impairs progenitor function and sabotages the regenerative niche. Enhancing mitophagy (e.g., with Urolithin A or genetic PINK1 overexpression) will break this loop, reduce senescence, and restore regenerative capacity.

Mechanistic Rationale

1. Mitochondrial stress as the primary trigger – Aged myocardium shows increased ROS and mtDNA damage in stem‑like cells [https://www.nhlbi.nih.gov/events/2026/inflammation-associated-hsc-senescence-and-dysfunction-blood-and-cardiovascular]. In c‑Kit+ progenitors, ROS can oxidize cardiolipin, leading to NLRP3 inflammasome assembly.
2. NLRP3‑driven SASP – Active NLRP3 cleaves caspase‑1, processing IL‑1β and IL‑18, which reinforce senescence via NF‑κB signaling and sustain p16INK4A expression [https://doi.org/10.1111/acel.12931].
3. Mitophagy as a brake – PINK1/PARKIN‑mediated mitophagy removes damaged mitochondria, lowering ROS and NLRP3 activation. Evidence from hematopoietic stem cells shows that boosting mitophagy rescues senescence [https://www.nhlbi.nih.gov/events/2026/inflammation-associated-hsc-senescence-and-dysfunction-blood-and-cardiovascular]; we extrapolate this to cardiac progenitors.
4. Feedback amplification – Senescent c‑Kit+ cells secrete IL‑1α and TGF‑β that further impair mitochondrial function in neighboring progenitors, expanding the senes

Hypothesis

Enlarged hydroxyapatite (HA) crystals in aged bone sequester undercarboxylated osteocalcin (ucOC), preventing its recycling by vitamin K–dependent γ‑carboxylation and impairing osteocyte mechanosensing. This creates a self‑reinforcing loop: trapped ucOC reduces HA‑binding competence of osteocalcin, further diminishing mineralization regulation, while altered HA size modifies pericellular fluid flow, lowering shear stress on osteocytes and increasing sclerostin expression.

Mechanistic Rationale

1. Cryo‑EM shows VKGC carboxylates osteocalcin in a strict order; low vitamin K yields mono‑ or di‑carboxylated forms that have reduced affinity for HA.
2. Larger HA crystals present a lower surface‑to‑volume ratio and altered charge distribution, creating nucleation sites that can adsorb ucOC via electrostatic interactions.
3. Osteoclast‑mediated decarboxylation releases OC into the resorption lacuna; ucOC that remains bound to HA is inaccessible to circulating vitamin K2, blocking reconversion.
4. Accumulated ucOC in the matrix competes with carboxylated OC for HA binding sites, weakening the feedback that normally limits crystal growth.
5. Changes in HA size and matrix composition alter interstitial fluid dynamics, reducing the fluid shear stress sensed by osteocyte processes; low shear stress upregulates sclerostin, suppressing Wnt‑driven bone formation.
6. Elevated sclerostin further decreases osteoblast activity, perpetuating low vitamin K recycling and ucOC accumulation.

Predictions

• In vitamin K‑insufficient aged mice, histomorphometry will show higher ucOC

Hypothesis

Aging neurons release exosomes whose cargo (specific miRNAs and metabolic enzymes) encodes a real‑time readout of cellular fitness; microglia survey these extracellular vesicles and initiate phagocytosis of neurons whose exosomal signature falls below a threshold, thereby executing an activity‑dependent pruning program that optimizes brain energy use.

Mechanistic Model

• Neuron‑derived exosome biogenesis is coupled to mitochondrial output and synaptic vesicle cycling. High ATP production and robust synaptic activity favor loading of miR‑124‑3p, miR‑132, and the enzyme COX4I1 into exosomes, while low activity shifts sorting toward miR‑155, miR‑146a, and increased phosphatidylserine exposure.
• Microglial sensing occurs via surface receptors that bind exosomal RNA‑binding proteins (e.g., hnRNPA2B1) and recognize altered miRNA ratios. A low‑fitness signature triggers downstream SYK‑dependent phagocytic signaling, whereas a high‑fitness signal engages CD47‑SIRPα pathways that inhibit engulfment.
• Feedback loop – removed neurons reduce local excitatory drive, lowering network-wide energy demand and allowing remaining circuits to operate more efficiently per watt, consistent with observed shifts in E/I balance during aging.

Testable Predictions

1. Correlation – In aged mouse hippocampus, exosomes isolated from CSF will show a significant increase in miR‑155/miR‑146a and a decrease in miR‑124‑3p relative to young animals; this shift will correlate spatially with areas of heightened microglial C1q deposition.
2. Loss‑of‑function – Conditio

Hypothesis

In aging vocal folds, progressive collagen I/III deposition and elastin loss increase lamina propria stiffness, which preferentially augments mechanical load on the lateral thyroarytenoid (TA) region. This altered mechanical environment dysregulates integrin‑FAK‑YAP/TAZ signaling in TA satellite cells, leading to aberrant calcium channel (CACNA1S) expression and impaired regenerative capacity. The lateral TA, already subjected to higher baseline tension during phonation, reaches a stiffness threshold that suppresses myogenic differentiation, explaining its disproportionate atrophy.

Mechanistic Rationale

1. ECM stiffness as a mechanotransduction cue – Elevated collagen cross‑linking raises tissue Young’s modulus, activating integrin β1‑FAK pathways that normally promote proliferation but, when sustained, drive YAP/TAZ nuclear accumulation and a fibrotic transcriptional program (e.g., TGF‑β1, CTGF). In skeletal muscle, chronic YAP activation suppresses MyoD and diverts satellite cells toward a fibro‑adipogenic fate.
2. Regional tension gradient – Finite‑element models of vocal fold vibration show lateral TA experiences ~15‑20 % higher cyclic strain than medial TA due to asymmetrical collagen fiber orientation. Higher strain amplifies integrin signaling, creating a feed‑forward loop that further stiffens the ECM via fibroblast‑mediated collagen synthesis.
3. Calcium channel coupling – FAK activation can upregulate CACNA1S transcription via ERK‑dependent mechanisms, as observed in aging rat lamina propria. Excess L‑type calcium influx promotes calp

Hypothesis

Recurrent SF3B1, SRSF2, U2AF1, and ZRSR2 mutations increase the formation of splicing factor condensates that trap nascent transcripts, raising local intron‑retention noise until a concentration‑dependent entropy threshold is crossed; cells that exceed this threshold become selectively vulnerable to spliceosome inhibition because their splicing machinery is already operating near collapse.

Mechanistic basis

Mutant spliceosomes alter Pol II kinetics and promote R‑loop formation, which stabilizes intron‑containing RNAs [https://pmc.ncbi.nlm.nih.gov/articles/PMC12765304/]. These RNAs phase‑separate with splicing factors, enlarging nuclear condensates and concentrating mutant spliceosome complexes. As condensate volume grows, the effective concentration of aberrant splice sites rises, increasing the probability that any given transcript will be mis‑spliced (intron retention or cryptic splice‑site usage). When the fraction of retained introns surpasses a critical entropy value—estimated from single‑cell transcriptomic noise metrics—the cell experiences a loss of buffering capacity for essential genes (e.g., MAP3K7, RAD51) and shifts from adaptive splicing plasticity to deterministic splicing failure.

This model extends the observation that cancers with spliceosome mutations show dependency on compromised splicing machinery [https://pmc.ncbi.nlm.nih.gov/articles/PMC12709091/] by proposing that the dependency is not static but emerges only after condensate‑driven entropy exceeds a threshold. Below the threshold, cells tolerate isoform noise; above it, they rel

Hypothesis

In aged dermal fibroblasts, the selective autophagy machinery prioritizes HAS2-containing complexes for degradation based on ubiquitin chain topology and lysosomal calcium signaling, thereby shifting hyaluronan synthesis from high‑molecular‑weight (HMW) to low‑molecular‑weight (LMW) fragments. This hierarchical targeting is not a passive consequence of bulk autophagy decline but an active triage mechanism that determines the extracellular matrix’s signaling output and drives fibroblast senescence.

Mechanistic Rationale

1. Ubiquitin code determines cargo rank – HAS2 resides in Golgi‑derived vesicles that acquire K63‑linked ubiquitin chains under oxidative stress. The density of these chains, amplified by the E3 ligase TRAF6, confers high affinity for the autophagy adaptor p62/SQSTM1.
2. Lysosomal Ca2+‑TFEB axis sets the threshold – Lysosomal calcium release via MCOLN1 activates calcineurin, which dephosphorylates TFEB, promoting its nuclear translocation and transcription of selective autophagy genes (including p62 and LC3). In aged fibroblasts, basal lysosomal Ca2+ is elevated, lowering the ubiquitin‑density threshold required for HAS2 engulfment.
3. Hierarchy over bulk flux – When autophagy is induced (e.g., by serum starvation), cargos are sorted by a "ubiquitin score": K63‑chains > K48‑chains > non‑ubiquitinated. HAS2 complexes, stress‑induced and heavily K63‑ubiquitinated, outcompete other membrane proteins (e.g., integrin β1) for limited phagophore nucleation sites, leading to their preferential sequestration.
4. Feedback via LMW‑HA –

Hypothesis: Age‑related loss of riboflavin‑producing gut microbes silences MAIT cell surveillance, turning these innate‑like lymphocytes into a source of senescence‑associated secretory phenotype (SASP) that drives epithelial leak, systemic inflammation, and further immune aging. Restoring MR1 ligands breaks this loop and should delay multi‑organ decline.

**Mechanistic rationale

• With advancing age, the Firmicutes/Bacteroidetes ratio rises and butyrate‑makers like Barnesiella fall (PMC7374892, PMC10967378).
• Concurrently, dietary‑derived riboflavin and bacterial synthesis of 5‑(2‑oxopropylideneamino)‑6‑D‑ribitylaminouracil (5‑OP‑RU), the potent MR1 ligand, drop sharply (inferred from loss of Lactobacillus and Enterococcus strains that produce riboflavin).
• MAIT cells, which depend on MR1‑presented 5‑OP‑RU for tonic signaling, receive weaker stimulation, shift toward an exhausted phenotype, up‑regulate PD‑1 and CD57, and begin secreting IL‑6, TNF‑α and ROS (PMC11464129, Frontiers in Aging 2022).
• SASP factors from MAIT cells impair goblet‑cell mucus production and tight‑junction proteins, worsening LPS translocation (PMC10967378, PMC11464129).
• The ensuing inflammaging further fuels myeloid‑biased hematopoiesi

Hypothesis

In hematopoietic malignancies bearing heterozygous SF3B1, U2AF1, SRSF2, or ZRSR2 mutations, oncogenic potency is not dictated by a single misspliced transcript but by the quantitative burden of co‑transcriptional splicing errors that generate R‑loops. When the ratio of retained introns to total nascent transcripts exceeds a definable threshold, ATR‑Chk1 signaling becomes constitutively activated, creating a synthetic‑lethal dependence on ATR activity. Below this threshold, cells tolerate splicing noise; above it, genome instability drives clonal expansion and determines therapeutic response to ATR inhibitors.

Mechanistic rationale

Mutant spliceosomes alter the kinetic coupling between RNA polymerase II elongation and exon definition. Elevated Pol II elongation rates, already observed in cancer increased transcription rates, reduce the time window for spliceosome assembly, increasing the probability of intron retention, especially at GC‑rich promoters prone to R‑loop formation. Retained introns expose nascent RNA that can hybridize with the template DNA strand, stabilizing R‑loops. These structures impede replication fork progression, activate ATR‑Chk1, and promote a mutational phenotype that favors clonal expansion.

Thus, spliceosome mutations act as “entropy amplifiers”: they increase the stochastic production of aberrant transcripts, and the functional impact emerges only when the cumulative error load surpasses a biophysical limit that overwhelms RNA‑processing and DNA‑repair capacities.

Testa

Immune aging is not a passive by‑product of organismal senescence but an active engine that fuels colonic field cancerization.

Hypothesis

We propose that accumulating senescent immune cells in the aged colon secrete mitochondrial‑derived reactive oxygen species (mtROS) and the cytokine IL‑6, which together suppress DNA mismatch repair (MMR) in neighboring epithelial cells. This creates a mutagenic niche that expands pretumorigenic clones bearing TP53, KRAS or CDKN2A mutations, thereby accelerating field cancerization and raising colorectal cancer (CRC) risk independently of extrinsic carcinogens.

Mechanistic Model

1. Senescent immune cell accumulation – With age, hematopoietic stem cells shift toward myeloid bias, increasing the pool of senescent neutrophils and macrophages that exhibit a senescence‑associated secretory phenotype (SASP) rich in mtROS and IL‑6 {[synthetic_research_questions_feedback_2.md]}.
2. mtROS transfer – Senescent immune cells release mtROS via extracellular vesicles or through transient membrane pores, delivering oxidative bursts to adjacent crypt stem cells.
3. IL‑6 signaling – IL‑6 binds epithelial IL‑6R/gp130, activating STAT3 and inducing expression of microRNA‑155, which directly down‑regulates MLH1 and MSH2 mRNA {[https://www.aging-us.com/news-room/epigenetic-aging-markers-predict-colorectal-cancer-risk-in-postmenopausal-women]}.
4. MMR suppression – Combined mtROS‑induced oxidative DNA lesions and reduced MMR capacity increase microsatellite instability and point mutations, providing a selective advantage to clones with

Hypothesis

With advancing age, bone hydroxyapatite (HA) crystals grow larger and become more stoichiometrically stable, increasing their affinity for γ‑carboxylated osteocalcin (Gla‑OCN). This enhanced binding depletes the pool of undercarboxylated, bioactive osteocalcin (Glu‑OCN) that normally restrains HA crystal maturation and signals to osteocytes through the GPRC6A receptor. Consequently, reduced Glu‑OCN signaling diminishes osteocyte calcium responses to mechanical load, contributing to the loss of mechanosensitivity observed in aged bone.

Mechanistic Basis

• HA crystal surface chemistry: Larger, more crystalline HA presents fewer defects and a higher density of phosphate groups, which strengthens the electrostatic interaction with the Gla domains of OCN (see vitamin K‑dependent carboxylation enables HA binding).
• OCN carboxylation shift: Elevated vitamin K availability or reduced warfarin‑like activity in aging shifts the Gla‑OCN/Glu‑OCN ratio toward the matrix‑bound form, as suggested by the correlation between increased HA crystallinity and OCN sequestration (Ageing modulates ultrastructural composition of bone matrix).
• Loss of Glu‑OCN signaling: Glu‑OCN inhibits HA nucleation and crystal growth (OCN inhibits mineralization and regulates HA crystal maturation). When sequestered, this brake is released, permitting runaway HA maturation.
• **Impact on osteocy

...

Hypothesis

Telomere length does not merely count divisions; it reflects the mitochondrial NAD+/NADH redox state through a telomere‑associated redox capacitor that modulates shelterin complex stability and p53‑dependent senescence signaling.

Mechanistic Basis

Telomeric G‑quadruplex structures can bind and transfer electrons, acting as a redox‑sensitive capacitor. Changes in mitochondrial NAD+/NADH alter local ROS production, shifting the redox potential at telomeres. A more oxidized environment destabilizes G‑quadruplexes, reducing shelterin binding and exposing telomeric DNA to DDR kinases. This triggers p53 activation and repression of PGC-1α/β, suppressing mitochondrial biogenesis and amplifying ROS—a feedforward loop that couples telomere status to metabolic dysfunction independent of replication count.[https://pmc.ncbi.nlm.nih.gov/articles/PMC5492019/][https://pmc.ncbi.nlm.nih.gov/articles/PMC3741661/][https://pmc.ncbi.nlm.nih.gov/articles/PMC7411297/][https://pmc.ncbi.nlm.nih.gov/articles/PMC6451386/]

Testable Predictions

1. Elevating the mitochondrial NAD+/NADH ratio (e.g., with nicotinamide riboside) will decrease telomere‑associated ROS signaling, lower TIF formation, and reduce p53‑mediated senescence markers without altering telomere length.
2. Conversely, inducing mitochondrial NADH accumulation (via complex I inhibition) will increase telomere oxidation, increase shelterin dissociation, heighten DDR signaling, and accelerate senescence even when telomerase is active.
3. Mutating G‑quadruplex‑forming sequences in telomeric repeats will uncouple telo

Hypothesis: Real-time AI-driven feedback that prompts individualized behavioral corrections when wearable metrics deviate from personal baselines will produce measurable improvements in objective metabolic and cardiovascular health markers in healthy adults, whereas sham feedback will not.

Mechanistic basis: Continuous glucose monitors and HRV sensors capture high‑frequency physiological noise that reflects immediate autonomic and metabolic responses to meals, activity, and stress. Prior work shows heart‑rate monitoring is rated most useful for fitness and recovery insights【Heart rate monitoring rated most useful in 2023 surveys for fitness and recovery insights](https://www.jmir.org/2025/1/e56251/), yet no direct evidence links self‑tracking to objective health gains in healthy cohorts【No direct evidence exists that self-tracking improves objective health outcomes in healthy individuals; benefits remain largely perceptual, such as self-reported diet or sleep improvements](https://www.jmir.org/2021/9/e25171/). A 2021 systematic review of 67 quantified‑self studies urges n‑of‑1 designs to test how personal baselines and fluctuations affect health【A 2021 systematic review of 67 studies on quantified self-tracking explicitly calls for n-of-1 experimental designs to understand how personal baselines and fluctuations—wearables capture continuously—impact health](https://www.jmir.org/2021/9/e25171/)**, establishing the methodological gap this hypothesis addresses.

We propose that an AI algorithm, trained on each user’s baseline streams, detects statistically significant

Hypothesis

Core claim: Incorporating sex‑specific epigenetic edge weights derived from methylation clocks into GNNs will rescue ranking performance (Hits@n) while reducing confounding bias in aging drug‑target prediction.

Rationale

• Existing GNNs achieve high accuracy but near‑zero Hits@n on DDI benchmarks [1]
• This paradox stems from models learning static topology that conflates true causal edges with confounding paths driven by latent variables such as sex, immune status, and diet [2]
• Mendelian randomization shows that unmeasured regulators bias edge inference [3]
• Prior work attempts to add confounder nodes [4] but remains untested in aging contexts.

Mechanistic Insight

We hypothesize that age‑related DNA‑methylation changes act as dynamic modifiers of protein‑protein interactions, effectively re‑weighting edges in the interactome in a sex‑dependent manner. When these epigenetic states are ignored, GNNs treat a static graph as if it were i.i.d., causing the model to assign high scores to many nodes that are merely correlated with confounding factors rather than causal drivers of aging. By encoding sex‑stratified methylation quantitative trait loci (meQTL) scores as edge attributes, the GNN can learn to down‑weight spurious confounder‑driven connections and up‑weight those that persist across epigenetic states, thereby improving the ability to rank true positi