Research directions

Why do people respond differently to the same meal, and what can these differences reveal about metabolic health? We combine meal challenges, continuous monitoring, multi-omics, and computational models to understand this variation and inform personalized nutrition.

Individual responses to meals

Physiological factors (gastric emptying, glucoregulatory hormones, amylase) and lifestyle factors (diet, exercise, sleep) shaping glucose levels measured by continuous glucose monitoring.

What factors affect glycemic responses to meals? Glycemic responses to identical meals vary across individuals and associate with metabolic subtypes (e.g., insulin resistance and beta-cell dysfunction) [7, 8, 9]. We aim to dissect the contributing factors — metabolic health, diet, behavior, genetics, and the microbiome — using both controlled meal challenges and real-life meals, to identify targets for reducing risk of metabolic diseases such as diabetes. We do both statistical modeling and new CGM cohort collection.

Dynamic regulation of metabolism

Glucose, G1P, alanine, and ethanol time courses and the highlighted central energy metabolism network in N. crassa, beside microbial pathways affecting metabolic traits directly and through metabolite mediation. Time courses and network adapted from Wu et al., PLOS ONE (2022), Figures 2B and 5B.

How do we understand dynamics in metabolic process? Human metabolisms respond dynamically to everyday perturbations (e.g., meals) on short timescales, providing natural experiments for understanding biological regulation. We develop computational methods that integrate time-series omics data to uncover how glucose dynamics shape — and are shaped by — broader metabolic processes, toward mechanistic understanding of disease and its complications.

AI for precision nutrition

Conceptual schematic linking glucose dynamics and food logs to an AI model, with bidirectional cohort interaction and predictions of muscle insulin resistance, beta-cell function, incretin effect, and liver insulin resistance.

How can AI help data collection and analysis in nutrition science? Advances in AI enable both deeper physiological insight (via foundation models) and scalable data collection and analysis (via agentic frameworks). We build models that infer metabolic health states such as insulin resistance and automate the capture and interpretation of real-life dietary records.