Lussier Yves, M.D.

Professor & Executive Director, Center for Biomedical Informatics and Biostatistics, University of Arizona Health Sciences

Yves A. Lussier

医学博士，美国医学信息学学会会士，美国亚利桑那大学生物医学信息学和生物统计学中心执行总裁、健康科学副主席兼首席知识官。其是一位研究精准医疗、转化生物信息学和个人基因组的临床科学家，并在该领域内国际知名。作为浦肯野细胞的发现人之一，他是将受控医学词汇表组织成电子病历中的定向语义网络、笔式计算用于商用的先驱。

Novel precision clinical trials designs enabled by machine learning applied to genome dynamics across single subjects studies

Imagine if we could predict response to each therapy with an affordable and timely test. Unfortunately, very few molecular biomarkers are truly accurate. Indeed, the US Food and drug administration approves only a handful of biomarkers yearly. Yet, more than 5% of funded USA NIH extramural grants (5,000 grants/yr) involve biomarker development!
Most often, these grants propose that a single molecular measurement can provide guidance for diagnostic, prognosis and treatment.
This monomolecular biomarker model is particularly well-suited for monogenic disorders and their responses to therapy (e.g., imatinib’s success in treating Philadelphia chromosome-positive chronic myelogenous leukemia). However, it has become apparent over the last decade that common diseases are caused by numerous gene-by-gene and gene-by-environment interactions (i.e., multigenic complex diseases), and it follows that modeling prognosis and response to therapy for these conditions also involves systems of molecules of life. IN other words, is our community of researcher we doing the same thing over and over again (single molecular biomarkers) and expecting a different outcome (nearly none acceptable for clinical care)?
The talk will focus on the discovery of interacting molecular dynamics involved in the development of systems-level biomarkers (multimolecular biomarkers). Specifically, we will propose biological essays that determine a clinically-relevant response to identify responsive systems-level biomarkers (SLBs). Different from conventional classifiers derived from “static panels of molecular measurements “, SLBs are designed ab initio from systems biology responses and machine learning modeling of genome-scale network dynamics. I will demonstrate the application SLBs to the framework of genome-by-environment classifier of future hospitalization of asthmatic children designed from in vitro cellular responses to rhinovirus. I will further broaden the applicability of these methods to single-subject trials.

Recommended reading:
• A genome-by-environment interaction classifier for precision medicine: personal transcriptome response to rhinovirus identifies children prone to asthma exacerbations. J Am Med Inform Assoc, 2017 Nov 1;24(6):1116-1126. PMID: 29016970