Evaluation of microelectrode array data using Bayesian modeling as an approach to screening and prioritization for...

Area of Science:

• Neuroscience
• Toxicology
• Computational Biology

Background:

• High-throughput in vitro screening is crucial for assessing chemical toxicity.
• Microelectrode arrays (MEAs) offer a promising tool for neurotoxicity screening by measuring neuronal network activity.
• Efficient data analysis is essential for the reliability of high-throughput screening methods.

Purpose of the Study:

• To develop and evaluate a Bayesian data analysis approach for assessing chemical effects on neuronal network activity using MEAs.
• To determine the sensitivity and specificity of the Bayesian approach for hit detection in neurotoxicity screening.
• To investigate the potential of combining the Bayesian approach with existing methods for improved accuracy.

Main Methods:

• A Bayesian data analysis method was developed to analyze spike data from primary cortical neurons cultured on multi-well MEA plates.
• The method was applied to a training set of 30 chemicals, analyzing spike counts from 64 microelectrodes per well.
• The Bayesian approach was evaluated independently and in combination with a weighted mean firing rate method.

Main Results:

• The Bayesian approach achieved approximately 74% sensitivity and 100% specificity in detecting neurotoxic chemicals within the training set.
• Combining the Bayesian approach with a weighted mean firing rate method significantly improved performance, yielding approximately 96% sensitivity and 100% specificity.
• The study demonstrated the utility of the Bayesian method for robust hit detection in MEA-based neurotoxicity screening.

Conclusions:

• A novel Bayesian data analysis framework provides an effective and reliable method for analyzing MEA data in neurotoxicity screening.
• The combined Bayesian and weighted mean firing rate approach offers a highly sensitive and specific tool for identifying neurotoxic compounds.
• This research supports the use of neuronal networks on MEAs as a valuable platform for chemical safety assessment and neurotoxicity testing.