2017/10/03 by Ritabrata Dutta, Dutta, Ritabrata, Bastien Chopard +9
Mathematics · #Statistical Methods and Bayesian Inference #Markov Chains and Monte Carlo Methods #Statistical Methods in Clinical Trials
paper · pdf · doi:10.48550/arxiv.1710.01054
Recent studies show the existing clinical tests to detect\nCardio/cerebrovascular diseases (CVD) are ineffectual as they do not consider\ndifferent stages of platelet activation or the molecular dynamics involved in\nplatelet interactions. Further they are also incapable to consider\ninter-individual variability. A physical description of platelets deposition\nwas introduced recently in Chopard et. al. [2017], by integrating fundamental\nunderstandings of how platelets interact in a numerical model, parameterized by\nfive parameters. These parameters specify the deposition process and are\nrelevant for a biomedical understanding of the phenomena. One of the main\nintuition is that these parameters are precisely the information needed for a\npathological test identifying CVD captured and that they capture the\ninter-individual variability. Following this intuition, here we devise a\nBayesian inferential scheme for estimation of these parameters. As the\nlikelihood function of the numerical model is intractable due to the complex\nstochastic nature of the model, we use a likelihood-free inference scheme\napproximate Bayesian computation (ABC) to calibrate the parameters in a\ndata-driven manner. As ABC requires the generation of many pseudo-data by\nexpensive simulation runs, we use a high performance computing (HPC) framework\nfor ABC to make the inference possible for this model. We illustrate that our\nmean posterior prediction of platelet deposition pattern matches the\nexperimental dataset closely with a tight posterior prediction error margin for\na collective dataset of 7 volunteers. The present approach can be used to build\na new generation of personalized platelet functionality tests for CVD\ndetection, using numerical modeling of platelet deposition, Bayesian\nuncertainty quantification and High performance computing.\n