Subscription is FREE for qualified healthcare professionals in the US.
5 Articles in this Series
Diet “Cheat” Days Negatively Impact Recovery From Inflammatory Injury
Ketogenic Diet Prevents and Rescues Mechanical Allodynia Induced by a High Fat Diet
Prolactin Linked to Sex Differences in Migraine
Toward Automated Pain Intensity Estimation in Mice: Finding Structure in Complex Data
Video Interviews

Toward Automated Pain Intensity Estimation in Mice: Finding Structure in Complex Data

Presentation by Jordan McCall, PhD

Neuropathic pain, which affects up to 10% of the world’s population, is one of the most severe and difficult forms of pain to treat.1 It is a common feature of a number of health conditions including diabetes, HIV, and cancer.

A major limiting factor in the development of novel treatments for neuropathic pain is that it is generally characterized by spontaneous pain occuring in the absence of a stimulus, which is difficult to study in animal models. New research presented at this year’s meeting describes the beginning stages of an approach that may someday circumvent this limitation.2

Researchers led by Jordan McCall, PhD, assistant professor at Washington University in St. Louis, used a method called persistent homology to automatically and blindly identify salient features of mouse behavioral data. This research was selected as one of the meeting’s most promising abstracts by the Scientific Program Committee and was presented as both a poster and a talk in a session on meeting highlights.

Stressed mice behave differently than non-stressed mice.

The researchers set out to identify unique hallmarks of mouse behavior using information hidden in the structure of the behavioral data itself. There are important nonlinear relationships imbedded in the structure of complex data sets that can be identified using topological data analysis methods, explained Dr. McCall. Persistent homology is a subfield of topological data analysis that can be used to tell if these nonlinear relationships are important.

First, the researchers determined whether persistent homology could even be applied to mouse behavioral data. They built a dataset to determine the homology of stress-induced anxiety behavior, as measured using the circular elevated zero maze. Stressed animals don’t fully explore the maze, and instead stay within the closed arm of the maze. Animals whose stress-induced anxiety has been blocked using the beta adrenergic antagonist propranolol explore the entire maze.

Using the variance of the behavioral data the researchers were able to blindly discriminate between the animals of each group, suggesting that persistent homology can be used to predict mouse behavioral phenotypes.

Next, the researchers applied this approach to a more complex behavioral assay of stress-induced anxiety, the open field test. A total of 80 mice completed the test immediately following 30 minutes of restraint stress. When the team ran a persistent homology analysis on the X-Y coordinates of the animals during the open field test, two groups emerged. One group was composed of 75% of the stressed animals, while the other was made up of 75% of the unstressed animals.

The researchers then repeated the experiment, this time administering propranolol or vehicle 30 minutes prior to the restraint stress, yielding a total of 4 groups: propranolol + stress; propranolol + no stress; vehicle + stress; and vehicle + no stress. As before, restraint stress induced anxiety in the mice, and this effect could be blocked with propranolol. Putting the data through the persistent homology analysis that was trained on the original open field test dataset yielded 2 groups: stress (made up of the vehicle + stress group) and no stress (comprised of the propranolol + stress, propranolol + no stress, and vehicle + no stress groups). The analysis predicted whether an animal was stressed or not with greater than 90% accuracy.

"We are very encouraged that we can use this relatively new branch of mathematics to make predictions from neuroscience data. So far, our results looking at animal behavior are promising and we are excited to extend these approaches into other areas of neuroscience and clinical applications," Dr. McCall told Practical Pain Management.

Most recently, the team has used their approach to examine pain models. Three weeks following bilateral hind paw injection of Complete Freund’s Adjuvant (CFA), a model of inflammatory pain, more than 90% of the animals (including both saline-treated and CFA-treated mice) were classified as no stress. This finding is consistent with previous research,3,4 showing that anxiety-like behavior may not develop in the CFA model until later time points.

In the spared nerve injury (SNI) model of neuropathic pain, all animals, including those that received a sham surgery, classified as stressed. Potential explanations for the stress classification of the sham mice are the shared housing with the SNI mice and the possible stress of the sham surgery.

“The end goal is to identify pain behaviors. The current analysis, which was built to examine stress-induced anxiety-like behavior in mice, has been somewhat successful and we think it will be necessary and important to build a similar approach specifically for pain-related behaviors. We are currently working towards addressing that very issue,” Dr. McCall concluded.



  1. van Hecke O, Austin SK, Khan RA, Smith BH, Torrance N. Neuropathic pain in the general population: a systematic review of epidemiological studies. Pain. 2014;155(4):654-662.
  2. McCall J, Hammarsten C, Bord A, Sheahan T, Golden J, Gereau R. Agnostic identification of mouse behavioral phenotypes. Poster presented at: Annual Meeting of the American Pain Society; May 17-20, 2017; Pittsburgh, PA. Poster #145.
  3. Narita M, Kaneko C, Miyoshi K, Nagumo Y, Kuzumaki N, Nakajima M, Nanjo K, Matsuzawa K, Yamazaki M, Suzuki T. Chronic pain induces anxiety with concomitant changes in opioidergic function in the amygdala. Neuropsychopharmacology. 2006;31(4):739-750.
  4. Parent AJ, Beaudet N, Beaudry H, Bergeron J, Bérubé P, Drolet G, Sarret P, Gendron L. Increased anxiety-like behaviors in rats experiencing chronic inflammatory pain. Behav Brain Res. 2012;229(1):160-167.
Next summary: Video Interviews
close X