Co-cultured sensory neuron classification using extracellular electrophysiology and machine learning approaches for enhancing analgesic screening

TL;DR

This paper explores using machine learning and electrophysiology to classify sensory neurons for better pain drug screening.

Contribution

A novel machine learning approach is proposed to classify responsive neurons in co-cultures for analgesic screening.

Findings

An RUS-boosted decision tree ensemble achieved an AUC-ROC of 0.877 in classifying nociceptors.

No single classifier outperformed others in accuracy, but the ensemble method showed strong performance.

Baseline neuronal activity was used effectively to train and validate classification models.

Abstract

Objective. Chronic pain affects over 20% of the adult population in the United States, posing a substantial personal as well as economic burden and contributing to the ongoing opioid crisis. Effective, non-addictive chronic pain treatments are urgently needed. Traditional drug discovery methods have failed to identify novel, non-addictive compounds, highlighting the need for alternative approaches such as phenotypic screening. Our lab developed a phenotypic screening assay using extracellular electrophysiological recordings from co-cultures of human induced pluripotent stem cell sensory neurons and glia. This study aimed to identify responsive neuronal subtypes within these presumptively heterogeneous cultures. Approach. We induced an inflammation-like state using tumor necrosis factor alpha and evaluated acute responses to nociceptor agonist capsaicin, which targets transient receptor potential vanilloid-1. By employing unsupervised learning, we labeled responsive cells based on changes in mean firing rates (MFR). We then used the labeled cells’ baseline activity to train and validate five classifiers. Main results. None of the classifiers outperformed the others in regards to accuracy. Nonetheless, an RUS-boosted ensemble of decision trees achieved an AUC-ROC of 0.877 classifying nociceptors in an unseen labeled culture. Significance. The notable accuracy suggests that machine learning techniques could be employed to enhance microelectrode array-based neuronal phenotypic assays as readouts (e.g. MFR) can be weighted based on target cell type (e.g. nociceptors).

Figures