Biomarkers of Aging Challenge

A collaborative initiative of Biomarkers of Aging Consortium, Methuselah Foundation, and Volo Foundation to accelerate the development of predictive biomarkers of aging models.

There is no clear scientific consensus on what biomarkers best measure the health impacts of aging and could be used for judging the effectiveness of medical or behavioral interventions. Without clear standards, the FDA and other regulatory bodies cannot judge the effectiveness of treatments, and the industry, in turn, struggles to bring interventions to the clinic.

Access to high quality omics datasets and disparate biomarker formulations are key roadblocks to reaching this goal. Moreover, there are disconnects between programming languages and methodologies favored by computational biologists and data scientists, which hinders the formation of transformative collaborations between such researchers.

We have generated a high-quality dataset featuring DNA methylation and aging outcome data including mortality/survival and age-related outcomes. Our dataset of 500 diverse individuals covers a wide range of chronological ages and is equally balanced between males and females. Importantly, our cohort also included individuals who have died since sample collection, allowing for the modeling of mortality risk. This dataset will form the basis for our Biomarkers of Aging Challenge, and will later be released as an open-access resource for the aging research community.

And in parallel, to enable innovation of the next generation of biomarkers of aging, we have built an open-source toolset called Biolearn. Biolearn is an unprecedented, one-stop, open-source platform for evaluation and validation of biomarkers of aging by curating and harmonizing large, high-quality omics and health datasets. We have built-in the ability to simultaneously calculate all currently available biomarkers of aging by harmonizing their formulations. Critically, Biolearn has been completely developed in Python to facilitate the entry of data scientists into the longevity space by providing standardized Python implementation of existing biomarkers to complement existing libraries in R.

The competition will be hosted on the Synapse platform and the winning models for each track will receive a cash prize.

Our competition has three consecutive tracks:

1. Prediction of chronological age.
2. Prediction of survival/mortality.
3. Prediction of multi-morbidity incidence.

Timeline

• 12/4/2023 - Team formation, open registration, and public comment phase begins.
• 3/1/2024 - Phase 1 - Chronological Age Prediction Challenge.
• Q3 2024 - Phase 2 - Longevity Prediction Challenge (mortality).
• November 1, 2024 - Winners celebration.
• TBD 2025 - Healthspan Prediction (multi-morbidity)