Another AI/ML tool for biological age?
This is quite interesting but it is still in peer-review status waiting for publication:
Wood T, Kelly C, Roberts M and Walsh B. An interpretable machine learning model of biological age [version 1; referees: 1 approved with reservations]. F1000Research 2019, 8:17
Note the quite critic review by Dr Alex Zhavoronkov of Insilico Medicine (see the comments following the paper) which already developed http://aging.ai/ and https://young.ai/#/ using an AI/ML approach (DNN) since already a couple of years. We discussed http://aging.ai/ previously in this Forum (see here, here and here)
Finally I wonder to which extent the SHAP (SHapley Additive exPlanations) plots might ease the determination of the relative weights importance of the Levine's Phenotypic Age as discussed here. Indeed it looks to me the technique does just that: "...SHAP plots of input markers. SHAP summary plots (Figure 2) were used to determine which markers have the greatest influence on predicted biological age...""
Should a free calculator based on this algorithms exist (or be developed?) I will certainly be curious to enter my data and check results.
Looking at the top 5 biomarkers the paper gives very similar results in terms of their relative importance to predicted age to those already published in 2016 by the team at Insilico Medicine which is the base of aging.ai v1.0: albumin, glucose, alkaline phosphatase, urea, and erythrocytes:
Putin E, Mamoshina P, Aliper A, et al. Deep biomarkers of human aging: Application of deep neural networks to biomarker development. Aging (Albany NY). 2016;8(5):1021-33.Reply