Human Phenotype Project 中多模态数字健康与组学数据的统计建模

The Human Phenotype Project (HPP) offers enormous opportunities to improve our understanding of human health by integrating medical imaging, biochemical measurements, omics data, and digital health profiles collected at multiple time points.

However, the multimodal, high-dimensional, and densely sampled nature of these data requires new analytical methods to ensure rigorous scientific inference and to fully exploit the information collected.

The goal of this project is to develop a new generation of statistical models for multimodal digital health and omics data and to apply them to clinically relevant problems, with the potential to generate important new insights into human health.

Objectives

1. Develop general statistical AI models for integrating digital health and omics data within the Human Phenotype Project.

2. Develop modality-specific statistical methods for data sources such as genome-wide association studies (GWAS) and microbiome data, with the aim of predicting functional digital health phenotypes.

3. Apply these models to fundus imaging data to investigate the genetic, metabolic, and demographic factors associated with the risk of retinopathy.

4. Conduct GWAS analyses to identify novel genetic variants associated with glucose metabolism and digital glycemic phenotypes.

5. Analyse microbiome data to identify microbial factors associated with glucose metabolism, with particular emphasis on glucose dynamics and postprandial responses.