Kevin Siswandi, a Master student in our group, gave a talk in the BioQuant Internal Seminar series on 16. April 2020. Due to the Corona virus pandemic the seminar was exclusively held online.
In his talk Kevin presented an AI project he has worked on last semester in our group with the title “Predicting the dynamics of biochemical systems from time-series multi-omics data“. Please see Kevin’s abstract below for further details.
In this project, we explore a data-driven modeling approach based on machine learning for predicting dynamics in biochemical systems from time-series data. Traditionally, dynamic modeling in systems biology often is a bottom-up process based on differential equations that relies on a hypothesis about the biological mechanism. However, it may take several years to map the pathway mechanisms in order to construct mathematical representations of the system. Moreover, bottom-up modeling approaches do not automatically scale in performance with more data and increase in complexity for larger systems. To solve these challenges, we blend machine learning, applied mathematics (numerical methods), and explainable AI to build a machine learning workflow that is not only superior in predictive performance on test systems, but also allows the extraction of mechanistic insights from the data. This work started as my internship at BioQuant and is now the topic of my Masters Thesis.”