Screening for dental pain using an automated face coding (AFC) software

Abstract

Objectives

This observational study evaluated the effectiveness of an Automated Face Coding (AFC) software in identifying facial expressions related to dental pain.

Methods

Fifty-seven participants (49.8 ± 17.1 years) with symptoms of dental pain were recruited. Participants self-reported their pain using a Visual Analog Scale (VAS) score and their faces were filmed using a smartphone. The video clips were exported to an AFC software, which analyzed the facial expressions. The analysis focused on detecting changes in facial expressions and emotional states. The analysis was performed at two timepoints, at baseline (on the first visit), and at post treatment recall when pain was alleviated (self-reported). Non-parametric tests were used for statistical analysis (p < 0.05).

Results

Significant reduction in pain levels was observed between the first visit and at the post treatment recall visit (mean VAS: baseline = 5.65 ± 2.08, recall = 0.40 ± 0.80; p < 0.001). No significant gender differences were observed in pain scores (p > 0.05). Significant differences in facial expressions between the two time points was not detected by the software (p > 0.05). Emotional parameters remained stable.

Conclusion

The findings of this study concluded that the current capability of the AFC software to detect changes in facial expressions specific to pain alleviation is limited, even though it can provide detailed analysis of facial muscle movements. Further research is needed to enhance the software's sensitivity to pain-related expressions and explore its integration with other diagnostic tools for improved patient care and treatment outcomes.

Clinical Significance Statement

The study explored the potential of AFC software in analyzing facial expressions for applications in screening and diagnosis of dental problems especially in non-communicative geriatric patients. While effective in monitoring facial movements, the software's current limitations in detecting pain-specific changes underscore the need for further advancements.