Embryonic.AI [title to be added]

July 5, 2016

Michael West,PhD., Co-CEO of BioTime, Inc., has announced the launch of a beta version of Embryonic.AI, an deep neural net (DNN) system for analyzing the embryonic state of human stem-cell and tissue samples using gene-expression data. The system is managed by BioTime subsidiary LifeMap Discovery, Inc.

“BioTime harnesses the largest collection of highest-quality gene expression data coming from scrupulously designed and controlled cell differentiation experiments we have seen to date,” said Alex Zhavoronkov, Ph.D., CEO of Insilico Medicine, Inc., developer of Embryonic.AI, which is now freely available to scientists for beta testing, via the Pharmaceutical Artificial Intelligence (Pharma.AI) division of Insilico Medicine.

EmbryonicAI’s architecture of multi-class deep neural networks (DNNs) has trained on thousands of samples of carefully selected embryonic stem cells, induced pluripotent stem cells, progenitor stem cells, adult stem cells, and adult cells to recognize the class and embryonic state of the sample and to predict their condition.

The unique samples were generated using standardized protocols of stem-cell and tissue cultivations by BioTime, Inc. and profiled (by gene expression of mRNA) on a single microarray platform.

The sample sets were augmented with carefully selected and manually curated data samples from public repositories to create training and testing sets. The dataset of gene expression samples was large enough to train a complex architecture of deep neural networks to work as a classifier and a predictor of the embryonic state.

“The study showed surprising results suggesting effectiveness in cross-species analysis in testing, using mouse data, Zhavoronkov said.

Embryonic.AI is a Baltimore-based bioinformatics company specializing in biomarker and drug development for aging and age-related diseases, with R&D resources in Belgium, Russia, and Poland.

The company has several research projects using Embryonic.AI in progress and planned. “They may transform our understanding of cancer and other diseases,” said Zhavoronkov. “Possible developments in reinforcement learning may also help navigate and control cellular differentiation states and in comparing multiple biopsies of patients’ tumor to search for cancer stem cells.”

The projects include organ engineering for drug testing, quality control of engineered human tissues to ensure that it closely resembles the expected results, analyzing the embryonic state of these tissues, and evaluating the effectiveness of many drugs in a high-throughput manner.”

Insilico Medicine has recently published several key papers on applying deep learning techniques to biomedical applications in influential peer-reviewed journals including “Deep learning applications for predicting pharmacological properties of drugs and drug repurposing using transcriptomic data” in “Molecular Pharmaceutics” and ACS publication. The paper received the Editors’ Choice Award from the American Chemical Society. “Applications of Deep Learning in Biomedicine” in also in Molecular Pharmaceutics and “Deep biomarkers of human aging: Application of deep neural networks to biomarker development” in Aging, one of the highest-impact journals in aging research. These studies were presented at the Machine Intelligence Summit Berlin on June 30th.

Insilico Medicine pursues internal drug-discovery programs in cancer, Parkinson’s, Alzheimer’s, sarcopenia and geroprotector discovery. Through its Pharma.AI division, the company provides advanced machine learning services to biotechnology, pharmaceutical and skin care companies. Since 2014, company scientists have published more thna 40 research papers in peer-reviewed journals and collaborated with over 150 academics, biotechnology and pharmaceutical companies worldwide.

Biotime, Inc. | Embryonic AI: Analysis of Embryonic State using Deep Neural Networks