Purpose <p>Recent studies showed that face recognition technology can be applied in clinical diagnosis. Blendshapes are one of the technologies to track facial movement. Oromotor functions involve the coordination and movement of swallowing muscles. It is hypothesized that blendshapes can track selected oromotor movements and classify between normal and impaired oromotor functions. This study aims to identify the relevant blendshapes to differentiate between normal and impaired oromotor functions.</p> Method <p>A total of 88 participants were recruited to carry out four oromotor tasks. All participants were instructed to conduct the tasks at their typical performance and simulate mildly and severely impaired oromotor functions. Movements were captured with an iPad Pro embedded with the TrueDepth camera. Two speech-language pathologists rated all videos independently on participants’ performance.</p> Result <p>Results showed that a small set of 5–13 parameters were significantly associated with each of the oromotor tasks, with large effect size differences across performance levels. Pairwise comparison revealed a significant difference in relevant parameters between normal and impaired oromotor functions.</p> Conclusion <p>This study supports the use of automated face recognition technology to track oromotor movements in healthy adults and identified corresponding blendshapes. Further testing in individuals with oromotor impairment is recommended.</p>