Biomedical research as well as the evaluation of the medical device at the Clinical Research Unit (algorithmic, data scientist, machine learning on Biomedical Big data, survival analysis, and experimental design in clinical trials research).

Biomedical research as well as the evaluation of the medical device at the Clinical Research Unit (algorithmics, data scientist, machine learning on Biomedical Big data).

I used my math and statistics know-how on problems in mathematical genetics. The first project here was to use data from genotyping arrays (mainly 50 K SNP) on segregating populations to infer which markers are involved in genomic structural variations. The idea behind this project is to usegenome-wide profiles of allele frequencies restricted to well-chosen individuals. Ideveloped the method in R and applied it to several datasets. I also provided validationsusing simulations to compare to the experimental data. This work is published in a high ranked journal and the R package available on R cran.My second project concerns the calculation of the probabilities of all possible multilocus genotypes arising in recombinant inbred lines of the « SIB » type. This had never been done for more than 3 loci, but I was able to tackle this difficult problem and very recently devised an approach that combines both analytical and numerical methods. My tricks now allow us to treat systems with up to 8 loci, a real quantum leap in performance. This work also is published in a high ranked journal and the code available online for use.

My thesis entitled <>.First, in this dissertation, we show that Bayesian networks classifiers are very accurate models when compared to other classical machine learning methods. We combine such approaches together with feature selection and discretisation to show that such a combination gives rise to powerful classifiers. The application to Epilepsy type prediction based on PET scan dataconfirms the efficiency of our approach. This work was published in ICMLA conference, CA, USA, 2017.Second, we also consider modelling interaction between a set of variables in the context of time series and high dimension. We suggest two approaches; the first is similar to the neighbourhood lasso where the lasso model is replaced by Support Vector Machines (SVMs); the second is a restricted Bayesian network for time series. This work was accepted in ICMLA conference, Cancun, Mexico, 2018.