A high-throughput machine vision-based univariate scale for pain and analgesia in mice

Abstract

Treatment of acute and chronic pain represent a widespread clinical challenge with poor therapeutic options. While rodents are an invaluable model to study pain, scoring nociceptive responses in clinically relevant paradigms and at high-throughput remains an unmet challenge. Therefore, there is a need for automated, high-throughput methods that sensitively and accurately assess pain and analgesia. Such objective and scalable technologies will enable the discovery of novel analgesics and yield mechanistic insights into the neural and genetic mechanisms of pain. Here, we adopt the open field arena to build a univariate scale for the formalin injection model of inflammatory pain by using a machine learning approach that incorporates 82 behavioral features. This tool outperforms traditional measures of licking and shaking in detection of formalin dose, and was validated using 4 diverse mouse strains. We also detected previously unreported differences in formalin induced nocifensive behaviors that were strain and sex specific. This model also reliably identifies morphine induced antinociception. This novel, sensitive, and inexpensive tool provides a method for quantifying voluntary nociceptive responses to facilitate genetic mapping and analgesic compound screening in a high throughput manner.

Visual detection of seizures in mice using supervised machine learning

“…Overall, all seizure associated behavior classifiers had good performance based on F β ( Table 2 ). Next, we extracted summary statistics for each behavior, including total behavior time, proportion of behavior time in trial, number of behavioral bouts, and average length of behavioral bouts similar to our previous applications ([ 27 ], fully described in Table S2 ). Combined with previously used open field features, we used 38 total features.…”

“…A univariate seizure scale offers several advantages: 1) easy to construct once the model is fit; 2) it is highly interpretable; and 3) it can easily be implemented with new data. We have applied it previously for frailty indexing and pain scoring [ 25 , 27 ].…”

“…We use a neural network trained to produce a segmentation mask of the mouse to produce an ellipse fit of the mouse at each frame as well as a mouse track. JABS is a versatile open source system, which we have used for gait and posture measures [ 29 ], grooming [ 24 ], biological age [ 25 ], and pain states [ 27 ].…”

“…These combine machine learning and computer vision (machine vision) and are often referred to as "artificial intelligence (AI)" [ 19 , 20 , 21 , 22 ]. We have successfully applied these methods for tracking [ 23 ], action detection [ 24 ], frailty [ 25 ], nociception [ 26 , 27 ], sleep [ 28 ], and gait and posture analysis [ 29 ]. Furthermore, we have built a shared open platform for end-to-end behavior annotation [ 30 ].…”

“…Thus, the most robust changes observed using unsupervised methods are interictal rather than ictal events. These findings, combined with our prior work utilizing supervised and heuristic classification of rodent behaviors, such as grooming [ 24 ], and combining multiple behaviors to score higher-level constructs such as frailty [ 25 ] and nociception [ 27 ] strongly suggest that the behavioral manifestations of frank seizures should be robustly detectable from video data, but no studies have addressed automatically scoring seizure severity per se in animal models. This is a significant gap because long-term monitoring of large cohorts may be critical for future pre-clinical studies of epilepsy.…”