Proposal Description

Advances in technology have facilitated high-throughput molecular profiling of many complex diseases in large-scale epidemiological studies. As costs decline, the number of studies collecting multiple modalities, simultaneously, only stands to grow. These studies have expanded our understanding of the genomic, transcriptomic, metabolomic, proteomic, microbiomic and epigenetic underpinnings of human health and disease, as well as the interplay between molecular layers. Although, there remains significant heterogeneity and limited understanding of directionality in findings. Accordingly, open areas of methodological and computational research include algorithmic fairness for disease prediction, causal discovery, and multimodal integration. Due to the inherent characteristics of the data, including its sequencing depth constraints, high dimensionality, and sparsity, many standard statistical approaches exhibit poor performance in omics data. Accordingly, statistical models tailored to each modality are needed to gain better biological insights. Motivated by the challenges in these open areas of research, the four speakers in this session will give presentations on these topics and discuss several novel statistical, machine learning, and artificial intelligence methods that narrow the existing methodological and computational gap. The statistical and AI developments discussed in this session act synergistically with research in medicine, public health, and biology to better understand the molecular underpinnings of complex diseases.