AI Longevity Project Opportunities

Bottom Line First (Project Judgment for Citation)

Conclusion 1 — pyaging is the only aging clock library today that meets the bar for a test run 🔬

Source: lucascamillomd/pyaging , GitHub Trending, most recent activity 2026-11-10. The repo describes itself as a GPU-optimized aging clock collection, primarily written in Jupyter Notebook, with 134 stars — niche but with a clear aging biomarker focus. License, data entry points, and exact clock count all need verification; README claims can’t replace an independent test run. Good fit for content creators or indie devs with a Python environment who want to do tutorial breakdowns, but not for personal biological age diagnosis or any health claims.

Conclusion 2 — TrajLift is a research lead in the Alzheimer’s early detection space, but there’s no testable asset yet

Source: TrajLift paper , Papers.cool q-bio, updated 2026-11-09, original publication date unknown. TrajLift uses a cohort of up to 746 people to build a spectral framework model on oral memory recall trajectories — it’s at the methodology pre-research stage; no public code repo, no data entry point. Good as material for content teams covering “AI cognitive detection research updates,” but not as a basis for products or detection tools.

Today’s Priority Project

lucascamillomd/pyaging — Python Aging Clock Collection

pyaging is the only project in today’s material that checks all three boxes: clear direction (aging clocks) + repo entry point + recent activity ✅. 134 stars is on the small side, but showing up on GitHub Trending means people are paying attention, and the Jupyter Notebook format is friendly for content creators doing breakdowns.

• Evidence source: https://github.com/lucascamillomd/pyaging , GitHub Trending, most recent activity 2026-11-10T04:34:50Z (not creation or release date)
• Credibility: Medium (repo exists and is recently active, but functionality, dependencies, and data entry points all need verification)
• Target audience: Longevity content creators, indie developers, research learners interested in aging clocks
• What it solves: pyaging bundles multiple aging clock models based on DNA methylation or other biomarkers into a single Python library. The repo claims GPU acceleration support, lowering the barrier for researchers who’d otherwise have to implement each clock individually.
• Test run barrier: Medium. Key unknowns: whether GPU is required, data entry points (whether reference methylation datasets are public or you need to bring your own), and dependency versions (Jupyter Notebook projects commonly have compatibility issues) — all need verification.
• License & data: TBD. Once in the repo, check the LICENSE file and README data source notes; don’t assume commercial use is viable until verified.
• Maintenance cost: Medium. Jupyter Notebook format carries a higher maintenance overhead (version drift, kernel dependencies), data updates depend on upstream methylation datasets, and single-maintainer risk is real.
• What you can build with it: Test run tutorial / content breakdown / tool navigation entry / resource library project card
• Deliverables: README test run notes (documenting install steps, errors, and pass/fail) / comparison table (listing supported aging clock names, input data types, GPU requirements) / resource library entry
• Minimum cost verification: Today you only need: clone the repo → read the README → list dependencies → compile the list of supported clock names → screenshot the README structure. No need to run the full pipeline.
• Success criteria: Within 24 hours, being able to list the clock names the repo supports, confirm the license type, and document the main dependencies = preliminary validation passed.
• Stop conditions: Pause further work if the license is non-commercial and doesn’t fit content creation needs, data entry requires payment or strict authorization, or dependency installation keeps failing in a standard environment.
• Secondary development path: After confirming the license, you could build a “mainstream aging clock comparison navigation page” (table format), then expand it into a weekly column “This Week in Aging Clock Research,” and eventually a lightweight web demo (upload methylation data CSV, get clock readings back) — would need someone with a bioinformatics background involved.
• Post-sale / compliance risk: Medium
• Compliance boundary: Any content must be framed as “research tool observation” and “biomarker research methodology introduction” — never as “measure your biological age,” “anti-aging efficacy assessment,” or any personal health diagnostic framing. Output values are research references, not clinical conclusions.
• Today’s minimum action: Open the repo homepage, read the README, and note: ① which clock names are supported ② whether a LICENSE file exists ③ where the dependency install instructions are ④ whether example data is available. Summarize into 3 bullet points and save to the resource library.

Directions for Secondary Development

Direction 1 — Aging Clock Comparison Table (based on pyaging + public literature) 📊

Multiple aging clocks (Horvath, Hannum, DunedinPACE, etc.) are scattered across different repos and papers, and there’s no consolidated comparison resource aimed at non-specialist readers.

• Deliverable: A comparison table listing clock name, input data type (blood/methylation/MRI), publication year, main validation cohort size, and limitations; paired with a “What aging clocks are, what they can do, and what they can’t” explainer note.
• Minimum cost verification: Today, just use the pyaging README + DunedinPACNI news material ( https://www.news-medical.net/news/20261109/Could-a-single-brain-scan-reveal-how-quickly-your-body-is-aging.aspx ) to compile 3–5 clock entries into a table. No code needed.
• Upgrade path: Publish the table as a searchable resource library page → add it to the weekly “New Clock Tracker This Month” → with a bioinformatics collaborator, expand into an online tool navigator.

Direction 2 — DunedinPACNI Research Material Compilation (content-focused) 🧠

The news material ( https://www.news-medical.net/news/20261109/Could-a-single-brain-scan-reveal-how-quickly-your-body-is-aging.aspx ) reports on DunedinPACNI (an MRI-derived brain aging metric) and its correlation with blood biomarkers across 746 people in two independent middle-aged cohorts, including 7–15 year longitudinal sub-group data. This is observational research, not an intervention trial — the conclusion is correlation, not causation.

• Deliverable: A popular science draft titled “Can MRI Measure Biological Age? What This Study Did and Didn’t Say,” focusing on study design, sample size, limitations, and conclusions that can’t be drawn.
• Minimum cost verification: Today, read the full news article, extract the study design (sample size, cohort type, follow-up years) and main limitations, and write up 3 bullet points for later use.
• Upgrade path: After accumulating 3–5 similar MRI brain aging studies, produce a “Cross-sectional Observation of Non-invasive Aging Detection Methods” feature.

Worth Watching

1 — biomarkersParkinson/ParaDigMa-Studio (Parkinson’s Wearable Digital Biomarkers) 👀

ParaDigMa-Studio is a Windows app that extracts Parkinson’s digital biomarkers from real wearable recordings, based on the ParaDigMa 1.1.2 toolkit. Source: https://github.com/biomarkersParkinson/ParaDigMa-Studio , GitHub Trending, most recent activity 2026-11-09, stars 6, Python.

Why hold off: stars are extremely low, Windows-only, license and data entry points need verification, and the ParaDigMa toolkit’s licensing is unknown. The target audience (Parkinson’s caregivers or medical professionals) has high compliance requirements, so writing tutorials or making recommendations before the license and clinical use boundaries are checked isn’t appropriate.

Tracking value: The wearable Parkinson’s digital biomarker direction aligns well with this radar’s aging/biomarker mainline. Re-evaluate once the repo’s activity picks up and the license is clarified.

2 — Zhaowado/biological-age-prediction-model

Zhaowado/biological-age-prediction-model has no description, and you can’t tell what data sources or input types it uses. Source: https://github.com/Zhaowado/biological-age-prediction-model , GitHub Trending, most recent activity 2026-11-03, stars 1, Python.

Why hold off: No description at all, data source and input type can’t be determined, stars are extremely low and activity is minimal. Until the README is read to confirm data entry points and license, test run feasibility can’t be assessed. Could be checked as a comparison entry after the pyaging tutorial is done.

Don’t Touch Today

1 — TrajLift Paper as Product or Detection Tool Basis ⛔

TrajLift is a methodology pre-proposal with no public code repo, no data entry point, and an unstated sample source in the material. Source: https://papers.cool/arxiv/2611.12197 , updated 2026-11-09, original publication date unknown. It can’t be packaged as an “AI voice detection for Alzheimer’s” product concept or used as an unvalidated personal health assessment tool.

2 — Mavrikaki-Lab Brain Aging Spatial Transcriptomics Code Repo ( https://github.com/Mavrikaki-Lab/Mavrikaki_Ra_Brain_aging_spatial_transcriptomics_human )

Mavrikaki-Lab’s brain aging spatial transcriptomics repo has 1 star, is in R, and the analysis code is tied to a specific study with no signs of general-purpose tooling. License and data entry points need verification; spatial transcriptomics data is typically large and access-restricted. The direction (human brain aging) is on-topic, but today there’s no viable path for content or toolification.

3 — Cancer/Cell Division & Lung Cancer TLS Research News ( https://www.genengnews.com/topics/cancer/how-cancer-associated-gene-triggers-cell-division-without-unpacking-the-genome/ , https://papers.cool/arxiv/2611.12855 )

Both pieces are tumor biology research, which falls outside this radar’s core focus on aging/longevity/dementia/biomarkers. The audience is cancer researchers, not this column’s target readers. Not included today.

Today’s Actions

• Test run first: Open https://github.com/lucascamillomd/pyaging , read the pyaging README, and note the list of supported clock names and dependency instructions. No need to run code.
• Write first: Compile 3 bullet points on “What aging clocks are” (based on the pyaging README + DunedinPACNI news material) and save to the content library as a topic candidate.
• Bookmark first: The pyaging repo; the original DunedinPACNI news article ( https://www.news-medical.net/news/20261109/Could-a-single-brain-scan-reveal-how-quickly-your-body-is-aging.aspx ).
• Make a table first: Draft an aging clock comparison table with at least pyaging’s self-described clock names + DunedinPACNI, noting input data types and limitations.
• Avoid today: Any direction that packages TrajLift, the spatial transcriptomics code, or the biological age prediction model as personal health assessment tools; don’t write a ParaDigMa-Studio tutorial before the license is verified.