AUTHOR=Platz Thomas , Pedersen Ann Louise , Bobe Stephanie 

TITLE=Feasibility, coverage, and inter-rater reliability of the assessment of therapeutic interaction by a humanoid robot providing arm rehabilitation to stroke survivors using the instrument THER-I-ACT

JOURNAL=Frontiers in Robotics and AI

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2023.1091283

DOI=10.3389/frobt.2023.1091283

ISSN=2296-9144

ABSTRACT=Objective. The instrument THERapy–Related InterACTion (THER-I-ACT) was developed to document therapeutic interaction comprehensively in the human therapist-patient setting. Here, we investigate whether the instrument can also reliably be used to characterize therapeutic interaction when a digital system with a humanoid robot as therapeutic assistant is used.
Methods. 
Participants and therapy: Seventeen stroke survivors receiving arm rehabilitation (i.e. arm basis training, ABT for moderate to severe arm paresis [n=9] or arm ability training, AAT for mild arm paresis [n=8]) using the digital therapy system E-BRAiN over a course of 9 sessions. 
Analysis of therapeutic interaction: A total of 34 therapy sessions were videotaped. All therapeutic interactions provided by the humanoid robot during the first and the last (9th) session of daily training were documented both in terms of their frequency and time used for that type of interaction using THER-I-ACT. Any additional therapeutic interaction spontaneously given by supervising staff or a human helper providing physical assistance (ABT only) was also documented. All ratings were performed by two trained independent raters.
Statistical analyses: Intraclass-correlation coefficients (ICC) were calculated for occurrence frequency and time used for each category of interaction observed.
Results. Therapeutic interaction could comprehensively be documented and was observed across the dimensions provision of information, feedback, and bond-related interactions. ICCs for the therapeutic interaction category assessments from 34 therapy sessions by two independent raters were high (ICC ≥ 0.90) for almost all categories of therapeutic interaction observed, both for frequency of occurrence and time used for categories of therapeutic interaction, and both for the therapeutic interaction performed by the robot, and – even though much less frequently observed – additional spontaneous therapeutic interaction by supervisory staff and a helper being present. ICC was similarly high for an overall subjective rating of concentration and engagement of patients (0.87).
Conclusions. Therapeutic interaction can comprehensively and reliably be documented by trained raters using the instrument THER-I-ACT not only in the traditional patient-therapist setting as previously shown, but also in a digital therapy setting with a humanoid robot as therapeutic agent, and for more complex therapeutic settings with more than one therapeutic agent being present.