Machine Learning to Detect Vocal Stereotypy: Improving Duration-Based Measures

Direct observation is a process central to behavior science, but its implementation occasionally requires hiring observers dedicated exclusively to data collection, which may increase the cost of providing services and conducting research. One potential solution to reduce resources necessary to conduct behavioral observation and measurement involves machine learning. Using data previously published by Dufour et al. (2020), we developed and tested novel models to automatically measure the duration of vocal stereotypy in eight children with autism. In addition to accuracy and the kappa statistic, we examined session-by-session correlations between values measured by machine learning and those recorded by a human observer. Nearly all our models produced high correlations (i.e., .90 or more) and resulted in better metrics than those reported by Dufour et al. (2020). The next step is for researchers to test the models on novel datasets to examine the generalizability of our findings.