eXplainable Artificial Intelligence (XAI) for the identification of biologically relevant gene expression patterns in longitudinal human studies, insights from obesity research

摘要: Until date, several machine learning approaches have been proposed for the dynamic modeling of temporal omics data. Although they yielded impressive results in terms model accuracy and predictive ability, most these applications are based on "Black-box" algorithms more interpretable models claimed by research community. The recent eXplainable Artificial Intelligence (XAI) revolution offers a solution this issue, were rule-based highly suitable explanatory purposes. further integration data mining process along with functional-annotation pathway analyses is an additional way towards biologically soundness models. In paper, we present novel XAI strategy (including pre-processing, knowledge-extraction functional validation) finding relevant sequential patterns from longitudinal human gene expression (GED). To illustrate performance our pipeline, work vivo GED collected within course long-term dietary intervention 57 subjects obesity (GSE77962). As validation populations, employ three independent datasets following same experimental design. result, validate primarily extracted prove goodness gene-gene relations. Our whole pipeline has gathered under open-source software could be easily extended to other applications.