> ## Content Index
> Fetch the complete content index at: https://www.coios.me/llms.txt
> Use this file to discover other available public pages before exploring further.

# A deep-learning senescence proteomic score predicted mortality and moved with exercise
- URL: https://www.coios.me/a-deep-learning-senescence-proteomic-score-predicted/
- Published: 2026-10-03T13:10:29.000Z
- Updated: 2026-10-03T17:29:29.000Z
- Description: A senescence-associated secretory phenotype score, built from UK Biobank plasma proteomics with a guided autoencoder-transformer model, independently predicted mortality and incident dementia, COPD, myocardial infarction and stroke, and changed over 18 months in a multimodal exercise trial.
- Author: Daniel Ryan
- Tags: Weighing data and models, AI in medicine, Slowing ageing itself, United Kingdom, #peer-reviewed, #new, #item, #source-2026-10

Researchers curated senescence-associated secretory proteins and trained a semi-supervised deep-learning model on UK Biobank Pharma Proteomics Project data to produce a single score, then tested it in an independent randomised trial cohort. The score was an independent predictor of all-cause mortality and of incident dementia, COPD, myocardial infarction and stroke, and its trajectory changed over 18 months under a multimodal exercise intervention.

*Why it is interesting: Takes a senescence measure from plasma biology to a population cohort and then to an intervention trial as a read-out.*

Source

[Aging Cell, 1 October 2026](https://doi.org/10.1111/acel.70737?ref=coios.me)

DOI

10.1111/acel.70737

Type

Peer-reviewed article

Design

Semi-supervised deep-learning (guided autoencoder with transformer) proteomic score developed in UK Biobank Pharma Proteomics Project, externally validated in an independent randomised trial cohort with 18-month follow-up

Verdict

New finding

Driver

[AI in medicine](https://www.coios.me/d-ai-medicine/)

Driver

[Slowing ageing itself](https://www.coios.me/d-ageing/)