Latent topic models to interpret patients' healthcare and genomic data
Yue Li - McGill University
Jan. 25, 2019, 2:30 p.m. - Jan. 25, 2019, 3:30 p.m.
Trottier 2120
Abstract:
Electronic health records (EHR) are rich heterogeneous collection of patient health information, whose broad adoption provides great opportunities for systematic health data mining. However, EHR data types and biased ascertainment impose computational challenges. Here, we present MixEHR, an unsupervised generative model integrating collaborative filtering and latent topic models, which jointly models the distributions of data observation bias and actual data using latent disease-topic distributions. We apply MixEHR on 12.8 million phenotypic observations from 46,000 patients at the intensive care units (ICU) Beth Israel Deaconess Medical Center, and use it to reveal latent disease topics, interpret EHR results, impute missing data, and predict mortality in ICU. Using both simulation and real data, we show that MixEHR outperforms previous methods and reveals meaningful multi-disease insights. Additionally, we further extend MixEHR to investigating the relationship between genetically-regulated transcriptional variation and phenotypic variation in a cohort of 9400 inner-city, low-socioeconomic-status, primarily-African-American patients with Post traumatic stress disorder (PTSD) at the Grady Memorial Hospital. We found several noteworthy enrichments for well-studied genes and tissues in specific PTSD-related topics across multiple tissues.
Speaker Bio:Yue Li’s research focuses on developing latent topic models to decipher, in a human-understandable manner, the etiology of diverse phenotypes including complex human diseases. By deriving interpretable and principled Bayesian learning frameworks, Yue seeks to answer some of the key questions in computational biology: what are the genetic predispositions, cell-type specificities, genomic regulatory elements, gene and pathway functions, and their interactions with environments that together give rise to the phenotypic diversity and their interdependency.
Yue obtained his PhD degree in Computer Science and Computational Biology at University of Toronto in 2014. Right after that, Yue became a postdoc at Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology. In 2019, Yue joined School of Computer Science at McGill as an assistant professor.
