On-Demand: Automated Human Phenotype Ontology (HPO) Terms Matching Using a Word2Vec Model Embedding

Presenter: Hamed Masnadi-Shirazi, Ph.D. | Principal Bioinformatics Scientist, Ambry Genetics

Exome analysis is time-intensive and strongly influenced by the similarity of patient phenotype with the phenotype associated with a gene. Assessment of clinical overlap can be automated using machine learning. This is accomplished by learning a Word2Vec model embedding of a publicly available HPO database to calculate gene phenotypic overlap with the patients’ clinical features.

During this webinar, the speaker will discuss:

An introduction to word embedding and similarity measures within natural language processing
An introduction to training and validating the Word2Vec model
A description of learning a Word2Vec model for the HPO and its application to the phenotype overlap problem

Dr. Masnadi-Shirazi earned a Ph.D. from the Electrical and Computer Engineering department at the University of California San Diego. His Ph.D. thesis was on designing optimized machine learning-based classification algorithms. Dr. Masnadi-Shirazi subsequently has 9+ years of experience in teaching and research in machine learning and has published in top-tier journals such as the Journal of Machine Learning Research, IEEE PAMI and conferences such as NIPS, ICML and CVPR. He has three years of experience in applying machine learning to bioinformatics and genetics problems at Ambry Genetics.

Automated Human Phenotype Ontology (HPO) Terms Matching Using a Word2Vec Model Embedding

Register Below for On-Demand Access:

HAMED MASNADI-SHIRAZI, PH.D.

Principal Bioinformatics Scientist, Ambry Genetics