Pilot Study to Discover Candidate Biomarkers for Autism based on Perception and Production of Facial Expressions

Overview

• This paper explores the potential use of facial expression as a biomarker for stratifying individuals with autism spectrum disorder (ASD) into subgroups.
• The study involved an online pilot experiment with 11 individuals diagnosed with ASD and 11 age- and gender-matched neurotypical (NT) participants.
• Participants were asked to recognize and mimic static and dynamic facial expressions displayed by 3D avatars, while their eye-tracking and facial movements were recorded.
• The researchers aimed to assess the validity of various constructs based on the expected responses of the NT group, and then identify the measures that could effectively discriminate between the ASD and NT groups.

Plain English Explanation

The researchers wanted to see if the way people with autism spectrum disorder (ASD) produce and perceive facial expressions could be used as a way to divide them into different subgroups. This could help with understanding the disorder better and developing more targeted treatments.

In this pilot study, the researchers had 11 people with ASD and 11 people without ASD (called "neurotypical" or NT) participate in an online experiment. The participants were shown 3D avatar faces making different expressions, and they had to try to recognize and mimic those expressions. The researchers used cameras to track the participants' eye movements and facial movements as they did this.

The researchers first checked to see if the measurements they were collecting from the NT group made sense based on what they would expect to see. Then, they used a statistical method to identify which of those measurements could be used to tell the ASD and NT groups apart.

The researchers found one potential eye-tracking measurement (how long people looked at the face when mimicking a "disgust" expression) and 14 other measurements related to facial movements that could be useful for dividing people with ASD into subgroups in the future. They also provided recommendations on how many people should be included in future studies on this topic.

Technical Explanation

The researchers designed an online pilot study to explore the potential use of facial expression production and perception as behavioral biomarkers for stratifying individuals with autism spectrum disorder (ASD).

The study involved 11 individuals diagnosed with ASD and 11 age- and gender-matched neurotypical (NT) participants. Participants were asked to recognize and mimic static and dynamic facial expressions displayed by 3D avatars. Their eye-tracking (ET) and facial video tracking (VT) data were collected, including measures of action unit (AU) activation and asymmetry from the Facial Action Coding System (FACS).

The researchers first assessed the construct validity of each dependent variable (DV) based on the expected response in the NT group. They then used the Boruta statistical method to identify DVs that were significant for discriminating between the ASD and NT groups.

The results of the study identified one candidate ET biomarker (percentage gaze duration to the face while mimicking static 'disgust' expression) and 14 additional DVs of interest for future study, including 4 ET DVs, 5 DVs related to VT AU activation, and 4 DVs related to AU asymmetry in VT. The researchers also provided sample size recommendations for future studies based on a power analysis.

Critical Analysis

The pilot nature of this study and its small sample size are limitations that the researchers acknowledge. The findings should be considered preliminary and in need of replication with larger, more diverse samples.

While the researchers used established methods for assessing construct validity and group discriminability, the specific choice of facial expressions and the use of 3D avatars rather than real human faces may have affected the ecological validity of the study.

Additionally, the researchers did not explore the clinical relevance or potential utility of the identified biomarker candidates. Further research is needed to understand how these measures relate to relevant functional outcomes and whether they can meaningfully stratify individuals with ASD.

Overall, this study provides a framework for ASD stratification biomarker discovery based on facial expression perception and production, but the findings should be interpreted cautiously and with an eye towards future validation and clinical translation.

Conclusion

This pilot study explored the potential use of facial expression perception and production as a means of identifying behavioral biomarkers that could stratify individuals with autism spectrum disorder (ASD) into meaningful subgroups. The researchers identified one candidate eye-tracking biomarker and several additional measures related to facial muscle activation and asymmetry that warrant further investigation.

While the findings are preliminary, this work represents an important step towards understanding the heterogeneity of ASD and developing more personalized, targeted interventions. Future research with larger, more diverse samples will be crucial to validate and refine these potential biomarkers, ultimately improving our ability to understand and support individuals on the autism spectrum.