Metabolic Modeling

What this research line covers

Metabolic modelling at the Metabolic Intelligence Lab means building predictive models of an individual metabolism from the data a person already generates every day — wearable sensors, dietary logs, body metrics — and using them to forecast how weight, energy balance and metabolic response will change. Three lines develop from it, each with its own page.

Metabolic Avatar

The digital replica of metabolism, in three generations: the Personalized Metabolic Avatar, calibrated on each user with gated recurrent units; the Generalized Metabolic Avatar, which needs no personal calibration and cut processing time from 12 seconds to 0.15; and the Quantum Metabolic Avatar, built on quantum echo state networks, which reduces prediction error by up to 76 per cent in noisy or data-scarce conditions.

Cardiovascular fitness

Detecting the aerobic threshold and other metabolic thresholds from the nonlinear dynamics of the heartbeat, measured during exercise, instead of assuming them from age-based formulas.

Chewing activity

Mastication as the first step of digestion: quantified with surface electromyography and related to satiety, digestive efficiency and metabolic response.

These models are what the devices of the lab run on: ReaCT for cardiovascular training and Byo-me for nutrition. The papers behind them are on the Publications page, with every DOI linked.