The rodent body language of ongoing pain

Abstract

To understand the nature of and to develop rational treatments for pathological pain, effective animal models capturing features of pathological pain are needed. As non-verbal, animals cannot self-report pain sensation and in consequence, pain perception must be inferred from behavioral signs or physiological markers. However, most laboratory models of pain-related behavior in current use are stimulus-evoked and/or reflex-based, targeting somatosensation. The limited current animal behavioral assays for spontaneous and ongoing pain are labor intensive, limiting throughput, and often suffer from subjectivity in selection and interpretation of actions indicative of pain sensation. For this dissertation, I developed a framework, Ethopain, for measuring the rodent body language of ongoing pain, engineered a machine vision and machine learning apparatus to implement Ethopain, and demonstrated that ongoing pain, and its relief by analgesics, can be captured in an objective and automated manner using Ethopain. Ethopain was further applied to building a pipeline for large scale in vivo analgesic efficacy validation, discovering novel pain related behavioral features with unsupervised learning, and building a probabilistic model to infer the shifting pain states in an individual mouse. In summary, this dissertation presents work that: 1. Gains a better understanding of the behavioral consequences of ongoing pain, with a focus on laboratory rodent voluntary behavior and 2. Develops new ways of measuring ongoing pain in rodents that are objective, sensitive, and scalable for genetic or pharmacological screening applications.