An Interpretable Machine Learning Model of Biological Age

March 22, 2019

When we launched the Blood Chemistry Calculator (BCC) in early 2018 we couldn’t have predicted the changes the software would undergo or the projects it would lead to. One such project has been researching and writing a scientific paper on the use of machine learning to predict and interpret biological age. The paper is currently in the peer review process on F1000Research, an open research publishing platform.

In this podcast, I talk with lead author Dr. Tommy Wood, MD, PhD, about the importance of knowing your biological age and understanding how it can be derived from basic blood chemistry markers.  Tommy and I discuss the peer-review process and the changes we’re making to the software as a result of the feedback that’s been provided. We also discuss the individual markers that have the greatest impact on biological age, and how you can get a free predicted age report.

Here’s the outline of this interview with Tommy Wood:

[00:00:58] Tommy got bit by a snake.

[00:02:38] Going to the doctor vs. changing lifestyle.

[00:03:32] Iatrogenic antibiotic injury.

[00:03:49] Antivenom: what it is, what it does and the side effects.

[00:06:49] Snake oral microbiota.

[00:10:23] Effects of antibiotics on gut.

[00:13:29] DUTCH (Dried Urine Test for Comprehensive Hormones).

[00:15:54] Our article: An interpretable machine model of biological age.

[00:17:15] Why is biological age important?

[00:19:12] Other tests of biological age; telomeres.

[00:20:31] Epigenetic testing.

[00:20:59] Effects of environment on epigenetic methylation; Studies: Nilsson, Emma, and Charlotte Ling. "DNA methylation links genetics, fetal environment, and an unhealthy lifestyle to the development of type 2 diabetes." Clinical epigenetics 9.1 (2017): 105; and Yet, Idil, et al. "Genetic and environmental impacts on DNA methylation levels in twins." Epigenomics 8.1 (2016): 105-117. Effects of lifestyle change on epigenetic methylation; Studies: Arpón, Ana, et al. "Impact of consuming extra-virgin olive oil or nuts within a Mediterranean diet on DNA methylation in peripheral white blood cells within the PREDIMED-Navarra randomized controlled trial: A role for dietary lipids." Nutrients 10.1 (2018): 15; and Delgado-Cruzata, Lissette, et al. "Dietary modifications, weight loss, and changes in metabolic markers affect global DNA methylation in Hispanic, African American, and Afro-Caribbean breast cancer survivors." The Journal of nutrition 145.4 (2015): 783-790.

[00:21:05] Epigenetic shifts and aging; Study: Pal, Sangita, and Jessica K. Tyler. "Epigenetics and aging." Science advances 2.7 (2016): e1600584.

[00:21:48] Insilico Medicine - Deep Biomarkers of Human Aging: aging.ai.

[00:22:46] Blood Chemistry Calculator (BCC).

[00:23:33] Find out your biological age with the free partial BCC report.

[00:24:04] How the biological age score is determined.

[00:28:13] Why we published the paper.

[00:28:40] Medscape article: Journal Editors on Peer Review, Paywalls, and Preprints.

[00:31:26] F1000Research.

[00:33:54] GitHub; XGBoost; Python.

[00:35:32] The reviewers for the peer review process: Alex Zhavoronkov and Peter Fedichev.

[00:39:10] Ideas that came out of the peer review process.

[00:42:49] Shapley Values and SHAP plots.

[00:43:51] Machine learning competition website: Kaggle.

[00:45:20] The most important blood markers for predicting biological age.

[00:48:02] Total cholesterol and BUN for predicting biological age.

[00:50:48] Nourish Balance Thrive on Patreon; NBT Forum.

Join the discussion on the NBT forum when you support us on Patreon.

© 2013-2024 nourishbalancethrive