Background

Digital twin platforms integrate multi-omics and longitudinal data to forecast organ-specific aging trajectories, achieving high accuracy in simulated cohorts [1]. These models leverage genomics, metabolomics, epigenetics and clinical biomarkers, yet they still show notable prediction errors when applied to real‑world, multimorbid individuals [4]. Multi‑omics analyses have also revealed gender‑specific patterns such as menopause‑accelerated aging [2], and longevity clinics are already translating comprehensive biomarker panels into actionable ‘Longevity Books’ [3].

Hypothesis

We propose that the primary source of prediction error in current digital twin models lies in unmeasured epigenetic alterations within tissue‑resident stem cell niches, which modulate systemic signaling via extracellular vesicle (EV)‑borne microRNAs. Incorporating niche‑specific EV‑miRNA profiles will significantly reduce the discrepancy between predicted and observed organ‑specific biological ages.

Mechanistic Rationale

Stem cell niches maintain tissue homeostasis through paracrine factors whose cargo is reshaped by niche‑specific DNA methylation and histone modifications. Age‑related epigenetic drift in these niches alters EV‑miRNA signatures, which in turn influence circulating metabolite and protein levels captured in multi‑omics panels but are not explicitly linked to organ‑age algorithms. When twins ignore this layer, they misattribute niche‑driven systemic changes to generic aging trajectories, inflating error especially in multimorbid individuals where niche perturbations are compounded.

Testable Predictions

1. In a longitudinal cohort of adults aged 60‑80, baseline plasma EV‑miRNA profiles will correlate with residuals between twin‑predicted organ ages (e.g., liver, brain) and clinically measured functional biomarkers after adjusting for age, sex, and comorbidities.
2. Adding a weighted EV‑miRNA term to the twin’s organ‑age algorithm will reduce mean absolute prediction error by at least 15% compared with the baseline model (p<0.01, paired t‑test).
3. Intervention‑specific EV‑miRNA shifts (e.g., after senolytic treatment) will precede measurable changes in organ‑specific biological age by ≥3 months, providing a leading indicator for twin‑guided therapy adjustment.

Methods (Outline)

• Recruit 500 participants from the Human Phenotype Project extension; collect plasma quarterly for 2 years.
• Isolate EVs, perform small‑RNA sequencing to quantify niche‑enriched miRNAs (identified via single‑cell ATAC‑seq of sorted stem cells from biopsy subsets).
• Compute organ‑specific biological ages using the existing physics‑informed neural network/LSTM twin [1].
• Calculate prediction residuals against gold‑standard measures (e.g., liver fibrosis via elastography, cognitive scores).
• Build multivariate models: baseline twin only vs. twin + EV‑miRNA latent factor (elastic net).
• Validate in a held‑out 20% set; assess calibration and discrimination.

Potential Implications

If validated, EV‑miRNA calibration would close a critical validation gap noted in current literature [4] by providing a mechanistically interpretable, minimally invasive biomarker that bridges molecular niche activity and systemic aging readouts. This could accelerate clinical deployment of digital twins in heterogeneous, multimorbid populations.