Integration through ICT
The data emanating from medical examinations, sample analysis and the IT developments will need to be integrated to form the virtual patient. ITFoM will develop the IT based methodology for the integration of these data. This methodology will include feedback loops allowing to further define and adapt medical practice, analytics and the IT infrastructure. Ultimately this will lead to an ever-improving, expandable and self-learning model of the human, ready to be used in various aspects of medicine and medicine-related research.
Modelling approaches in ITFoM
In the ramp-up phase, four different modelling approaches and combinations thereof will be developed and used to compute data from the two ITFoM use cases: metabolic syndrome and colon cancer.
Statistical analyses of predictive power will be used to design combination predictions that are most succesful.
The Watchmaker: mechanistic model of macromolecular processes in the human (based on kinetic rates and concentrations of molecules)
The Engineer: model based on representation of objects corresponding to specific cell/tissue types exchanging molecular information (via growth factors, hormones, nerve signals, drug concentration…) in systems of differential equations or petrinets.
The Mechanic: The Mechanic’s approach to integration is a coarse‑grained, mostly non‑molecular and multiscale model; it starts from the scale where the pathology is observed (typically organ or tissue scales) and then adds scales above, up to the whole body and below, down to the tissues and interactions with the cellular and subcellular components. It is rooted in a concept of space‑time, and thus frequently requires PDE systems and spatial fields as mathematical building blocks.
The Learner: This approach uses computational statistics and machine learning procedures and will be applied in combination with mechanistic information as constraint in the procedures.
The modelling approaches will be applied to the use cases Metabolic Syndrome and Colon cancer.
The Watchmaker approach will generate a network representing the transport (of substrates such as short‑chain fatty acids and glucose, and of xenobiotics), cell cycling, signal transduction (MAP kinase, Akt/IGF, TOR), gene expression and epigenetics, detoxification (NADPH, glutathione, pentose phosphate pathway), and mitochondrial apoptosis will be connected to a focus on oxygen‑dependent carbon and energy metabolism in colon epithelial cells and in microbial cells in the adjacent colon.
The Engineer’s aproach will model use cases and external data sets using available data to determine changes in function and concentration of the object on the biological networks relative to the object in the reference model. Where possible, predictions on treatment response in cancer patients will be validated on xenografts and ultimately in patient treatment.
The Mechanic’s approach will develop a continuous dynamic model from transformed colon to whole body in the context of metabolic syndrome and the metabolic function of the colon, as well as of colon cancer. Using a reference and a number of personalised versions, predictions will be made for individuals that differ substantially both in the context of metabolic syndrome and of the relation between inflammatory bowel disease (IBD) and cancer.
Inspirations for Biology-driven ICT
The integration results will be used to benchmark the computation efficiency of the modelling methodologies. This will feedback to the developers of hardware, software and computational methodologies to drive "biolgy-inspired" ICT.