Posted on 25 May 2016
Bioinformatics company Insilico Medicine has developed AgingAI, an online platform that can predict your age based on biomarkers in your blood
How was it developed?
Using data from the largest independent laboratory test provider in Eastern Europe, Invitro Laboratories, over a million samples were analysed to create a data set of over 60,000 samples. The Insilico medicine team then created a modular ensemble of 21 deep neural networks (DNNs) of different depth and structure. These networks can outperform other machine learning methods, and were able to identify key aging biomarkers in the samples. The resulting ensemble could actually correctly predict age in 80% of samples, which is a big step forward.
The biomarkers best able to predict chronological age were:
Glucose, albumin, alkaline phosphatase, urea and erythrocytes
“It is exciting to see the power of deep learning applied to potential aging biomarkers. The availability of such markers is an essential prerequisite for any future clinical trials to try to ameliorate the effects of human aging” – Charles Cantor
Why is this good news?
We really need simpler, more efficient ways to determine chronological age. Utilising blood test data in combination with an online platform like AgingAI could be a great way of measuring biomarker changes in clinical trials. If a therapy holds regenerative promise, then measuring changes like this with an AI could reveal whether it’s working or not. Insilico Medicine is currently working on many other applications of deep learning methods, across multiple fields including regenerative medicine.
“We decided to train an ensemble of deep neural networks on a very large number of simple inexpensive historical blood tests linked to age and sex and built a predictor, which is scalable and can include many other data types to build more comprehensive biomarkers of aging. Aging.AI can in principle be extended as a biomarker of biological aging that can be used to assess the efficacy of various therapies” – Poly Mamoshina
Read more at EurekAlert