Basic Science: Pain Models

Artificial Intelligence-Based Approaches to Preclinical Pain Assessment

Measurement of pain in non-human species in laboratory, clinical and domesticated environments is limited to surrogate outcome measures due to their inability to communicate, unlike humans. In this context, the interpretation of the pain state is based on the measurement of behavior as indicators of the impact of pain on various dimensions. Behavior of animals is dramatically influenced by the presence of humans, resulting in unpredictable biases during experiments. This fact, coupled with the increasing realization that the evaluation of complex behaviors may be more clinically relevant, has led researchers to explore ‘hands-off’ methodology to measure complex behaviors. The explosive growth and availability of high-quality video and computational technology is facilitating the development of automated or semi-automated methods to capture and analyze complex behaviors as surrogate measures of pain and analgesia. Attendees will meet and interact with experts across a diverse range of non-human species (flies, rodents, large animals) – experts who will explain, discuss and demonstrate the use of video and artificial intelligence (AI) in its application to the measurement of pain. AI-based automated video analysis is becoming increasingly available and utilized, and, as for any methodology, an awareness of its advantages and caveats will be essential for all scientists working with non-human species.
The AI-based approaches used will be described and demonstrated in the context of particular species but have very broad potential applications across species and pain models.

Read More

Automated Pain-Related Activity Tracking in Fruit Flies

There will be a video demonstration of how to measure and interpret nociception in drosophila and how such techniques can then be used to identify novel genes for forward translation. Different analysis techniques for monitoring and interpreting activity in single and group situations, will be demonstrated.

Read More

Automated Mouse Behavior and Pain-Related Behavior Tracking

Ms Maree focuses on building tools that leverage computer vision and deep learning to tackle problems in neuroscience and biology that would be otherwise intractable, including markerless motion capture to quantify animal behavior. The reflex limb withdrawal response as a measure of nociception has been a mainstay of quantitative measurement in preclinical pain research in subjects unable to describe pain. There have also been attempts in the literature to measure effects on activity and social behavior. There will be an explanation and video demonstration of how high-quality computer-aided vision and deep learning can be applied to different aspects of behavior of single and socially housed mice using software developed by the lab called SLEAP.

Read More

Automated Facial Image Evaluation

There will be a demonstration of the grimace scale used in small and large animal species, with a discussion of data collection and scoring. This will build into how the standard, human observer approaches, contrast with evolving automated, machine learning approaches, including a demonstration of automated image scoring.

Read More

Automated Analysis of Homecage Lid Interaction and Hanging Behaviour as a Measure of Pain in Mice

Most rodent studies of pain rely on the use of experimenter-evoked measures of pain and assess behavior under ethologically unnatural conditions, which limits the translational potential of preclinical research. We approached this problem by conducting an unbiased, prospective study of behavioral changes in mice within a natural homecage environment using conventional preclinical pain assays. We observed that cage lid hanging, a species-specific elective behavior, was the only homecage behavior reliably impacted by pain assays. Noxious stimuli reduced hanging behavior in an intensity-dependent manner, and the reduction in hanging could be rescued by analgesics. Finally, we developed an automated approach to assess hanging behavior. This demonstration will review the utility and limitations of studying homecage behaviors for the assessment of pain and describe technologies to automate the measurement of cage lid interaction.

Read More

Large Animal Automated Behavior Scoring using Machine Vision and 3D Cameras

Automated measurement of animal behavior is possible using machine vision systems. We have worked with a small Agri-tech company using 3D cameras on farm to weigh pigs, but also to measure the posture of their tail, and we are working on measuring a wider range of behaviors. We first showed that tail posture in a group became lower response to the onset of a painful vice behavior – tail biting. On commercial farms, low tail posture was affected by tail biting, but also lameness, and aggressive social behaviour, suggesting it could be useful as a wider indicator of animal wellbeing.

Read More

Automated Detection and Analysis of Pain-Related Behaviors in Rodents: Gait and Movement Analysis

The development of new analgesics is challenging, in part, due to the poor translational value of current rodent pain assessment tools and the difficulty and inefficiency of their use. Alternative technologies are needed that are better able to predict clinical effects of experimental analgesics and with high-throughput capability. Blackbox Bio’s technology addresses this unmet need with a system that permits automatic detection and scoring of voluntary pain-related behaviors in laboratory rodents. By leveraging machine learning and vision tools, the Palmreader™ technology reveals the innate “body language” of an animal recorded in a dark and observer-free environment from a bottom-up view. With this technology, the pose and plantar contact features of freely-behaving rodents are analyzed to reveal their internal state of pain and analgesia. These readouts are more sensitive to pharmacological intervention than conventional assays and can be readily differentiated from other pharmacologically-evoked behavioral features (Zhang, et al., Pain, 2022).

Read More