Uncertainty Quantification Driven Predictive Multi-Scale Model for Synthesis of Mycotoxins

Many toxic molds synthesize and release an array of poisons, termed mycotoxins that have an enormous impact on human health, agriculture and economy [1]. These molds contaminate our buildings, indoor air and crops, cause life threatening human and animal diseases and reduce agricultural output [2]. In order to design appropriate approaches to minimize the detrimental effects of these fungi, it is essential to develop diagnostic methodologies that can rapidly and accurately determine based on fungal strains and their growth patterns, the extent of mycotoxin mediated damage caused to the environment.Here we developed a novel multi-scale predictive mathematical model that could reliably estimate aflatoxin synthesis from growth features extracted from Aspergillusparasiticus, a well-characterized model for studying mycotoxin biosynthesis. We conducted acoustic imaging experiments to observe and extract the growth features from the biomass profiles of the growing Aspergillus colony growing on an aflatoxin-inducing solid growth medium. We employed the probability-based representation of uncertainty and used Bayes’ theorem to infer the uncertain parameters in our mathematical model using biomass observations of the colony at 24h (aflatoxin is not synthesized yet at this time-point) and 48 hours (aflatoxin synthesis occurs at peak levels). We demonstrate that our model could successfully predict with quantified uncertainties the levels of aflatoxin secreted to the environment by the fungus.