The benefits of caregivers implementing Pivotal Response Treatment (PRT) with children on the Autism spectrum is empirically supported in current Applied Behavior Analysis (ABA) research. Training caregivers in PRT practices involves providing instruction and feedback from trained professional clinicians. As part of the training and evaluation process, clinicians systematically score video probes of the caregivers implementing PRT in several categories, including if an instruction was given when the child was paying adequate attention to the caregiver. This paper examines how machine learning algorithms can be used to aid in classifying video probes. The primary focus of this research explored how attention can be automatically inferred through video processing. To accomplish this, a dataset was created using video probes from PRT sessions and used to train machine learning models. The ambiguity inherent in these videos provides a substantial set of challenges for training an intelligence feedback system.

Original languageEnglish (US)
Title of host publicationSmart Multimedia - 1st International Conference, ICSM 2018, Revised Selected Papers
EditorsStefano Berretti, Anup Basu
PublisherSpringer Verlag
Number of pages12
ISBN (Print)9783030043742
StatePublished - 2018
Event1st International Conference on Smart Multimedia, ICSM 2018 - Toulon, France
Duration: Aug 24 2018Aug 26 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11010 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other1st International Conference on Smart Multimedia, ICSM 2018


  • Attention detection
  • Autism spectrum disorder
  • Dyadic behavior
  • Pivotal response treatment
  • Untrimmed video

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Detecting attention in pivotal response treatment video probes'. Together they form a unique fingerprint.

Cite this