Abstract: A method of generating a metabolic digital twin of a subject for clinical decision support in relation to a medical condition, the method comprising: receiving data indicative of an extended metabolic map comprising nodes representing a plurality of metabolites and one or more non-metabolic parameters, and edges representing relationships between them; determining an extended stoichiometry matrix at least partly from the extended metabolic map, wherein coefficients in the extended stoichiometry matrix define quantitative relationships between abundances of the metabolites and/or values of the non-metabolic parameters; receiving data indicative of measurements of a sample of the subject, wherein the measurements comprise measurements of one or more of available metabolite concentrations for one or more of the plurality of metabolites, and measurements of one or more of the non-metabolic parameters; and optimizing, subject to one or more constraints, an objective function that depends on a product of the stoich