Deep-Ephys – Using machine learning for biomarker discovery in human iPSC neuronal networks
- Manuel Schröter, ETH Zürich
- Damine Roquiero, ETH Zürich
- Felix Franke, ETH Zürich
- Karsten Borgwardt, ETH Zürich
- Andreas Hierlemann, ETH Zürich
The primary scientific goal of this project is the development of biomarkers for the quantification of phenotypes of neuronal networks from MEA recordings. A biomarker, in this context, is a function that maps the spatial and dynamical features of neuronal networks (comprising up to thousands of neurons) onto a small set of numbers. These numbers, when extracted from cultures with different phenotypes (e.g., diseased vs. healthy), should be maximally discriminative for the phenotype. Such biomarkers would be of great importance, e.g., for drug screening: Does the application of a new drug to a diseased cell culture lead to a convergence of the biomarker to the healthy state? Furthermore, the goal is to bridge the technical gap between data acquisition and data analysis.