B0223
Title: Similarity-based multimodal regression
Authors: Russell Shinohara - University of Pennsylvania (United States)
Haochang Shou - University of Pennsylvania (United States)
Andrew Chen - University of Pennsylvania (United States) [presenting]
Abstract: To better understand complex human phenotypes, large-scale studies have increasingly collected multiple data modalities across domains such as imaging, genomics, and physical activity. The properties of each data type often differ substantially and necessitate either multiple separate analyses or extensive processing to obtain comparable features for a single analysis. For a single data modality, multivariate distance matrix regression provides a distance-based framework for regression involving a wide range of data types. However, no distance-based method exists to handle multiple types of data in a single analysis. We extend a distance-based regression model to propose similarity-based multimodal regression (SiMMR), which enables simultaneous regression of multiple modalities through their distance profiles. We demonstrate through both simulation, imaging studies, and longitudinal wearable device analyses that our proposed method can detect associations in multimodal data of differing properties and dimensionalities, even with modest sample sizes. We furthermore perform experiments to evaluate several different test statistics and provide recommendations for applying our method in a wide range of scenarios.
