RNN Models for Clinical Reverse Engineering & Predictive Outcome
Integration of different kinds of biomedical data as genomics and proteomics with clinical data, providing direct feedback to the health care system is a huge opportunity area. In the past I developed a time delayed recurrent neural network (TDRNN) optimized by a genetic algorithm to integrate time series with different time scale as cell signal transduction and gene regulatory networks. This allowed for reverse engineering of the system as well as for modelling and simulation. Now I am extending this platform to handle clinical “static layers” of information by a “rule based” integration that switches to a Hopfield like behavior for memory association. In this way closing the loop towards the desired integrated platform.
David Camacho's PhD work won, for the first time at the institution, the Intramurales Förderungsprogramm from the German Cancer Research Center (DKFZ) and is involved on a pending patent. His work as scientist includes leading the integrative analysis of “omics” with clinical (oncologic) consortium’s data and modelling/simulation in the Belgium´s Center for medical genetics at Gent. Additionally, he used his systems biology expertise in the field of small molecules screening to systematically correlate structure-pahrmaco dynamics, at two different CRO´s in Switzerland and England.