Robot-Assisted Intervention for children with special needs: A comparative assessment for autism screening

https://doi.org/10.1016/j.robot.2020.103484Get rights and content

Highlights

  • ASD risk factors associated with CwASD ability to manifest JA were identified in RMI.

  • CwASD group showed a different behavioral pattern in three autism signs.

  • CwASD did not perform better in the RMI; this does not mean that RMI are not suitable.

  • Robot was relevant, because it led to demonstrate differences between the two groups

Abstract

Despite the increment of researches related to Social Assistive Robotics (SAR), achieving a plausible Robot-Assisted Diagnosis (RAD) for Children with Autism Spectrum Disorders (CwASD) remains a considerable challenge to the clinical and robotics community. The work of specialists regarding ASD diagnosis is hard and labor-intensive due to the condition’s manifestations are inherently heterogeneous and makes the process more difficult. Besides, the aforementioned complexity may be the main reason for the slow progress in the development of SAR with diagnostic purposes. Thus, this work provides a comprehensive Robot-Assisted Intervention for CwASD showing the conditions in which a Robot-based approach can be useful to assess autism risk factors for an autism diagnosis purpose. The intervention scheme consists of an improved version of a multimodal environment for Robot-based intervention proposed in our previous work. More specifically, we compared the behavior of CwASD with that of children in a control group during a human/robot-mediated intervention while Joint Attention (JA) behaviors are elicited and analyzed. Through statistical data analysis, it was possible to identify that 17 out of 23 children of the CwASD group showed a different behavior pattern related to three characteristics of autism, which suggests that this pattern can be used to identify autism risk factors through Robot-based interventions.

Introduction

Currently Social Assistive Robotics (SAR) and Child–Robot Interaction (CRI) researches have shown prominent results that have aroused a lot of interest within the autism community Scassellati et al. [1]. However, achieving a plausible Robot-Assisted Diagnosis (RAD) for Children with Autism Spectrum Disorders (CwASD) remains a considerable challenge to the clinical and robotic community. Surprisingly, the view of assisting the autism diagnosis using robots is not new. In 2007, Scassellati [2] proposed that robot-assisted diagnosis would be one of the applications with the most significant potential for the autism community. However, there are few reported works regarding robot-based diagnostic tools. In contrast, many works regarding robot-assisted therapy (RAT) and an improved version called robot-enhanced therapy (RET) have been reported in [3]. Currently, the development of RAT applications continues increasing. An example of this trend is reflected in the reviews of Pennisi et al. [4], which reported 25 RAT from 2009 to 2016. Also, Begum et al. [5] reported a survey with 14 RAT studies between 2009 and 2014. However, no studies regarding assisted diagnosis tools were found in these reviews.

The slow progress in the development of SAR-based tools to help specialists in the diagnosis of autism may have been the cause of the lack of these studies [6]. In fact, the traditional diagnosis of autism is already stressful enough, given the high variability of signs exhibited for CwASD [7]. The traditional process of ASD diagnosis requires from the medical specialist to address behavioral assessments of the child’s development state in four domains, such as behavior excesses, communication, self-care, and social skills [8]. These assessments require responses to a large number of paper-based questionnaires, which makes many of them lengthy and labor-intensive, such as Checklist for Autism in Toddlers (CHAT) [9], and Gilliam Autism Rating Scale (GARS) [10]. In addition, the literature shows that there is a time gap among parents’ first concern about the child’s development impairments, their first medical evaluation, and the child’s age of confirmed diagnosis [11]. Furthermore, it has been identified that child care and education centers have more opportunities to recognize risk factors of ASD in children than pediatric surveillance system.

From a technical perspective, the lag in manifesting real benefits of robots to make autism screening as well as the modern assessment and diagnosis is due to the difficulty of finding mechanisms to analyze automatically children’s behaviors in naturalistic environments, elicit behaviors through few robot interventions and also generate clinically valid metrics that allow identifying behavioral patterns to confirm risk factors of ASD [2].

Although the concerns and objectives associated with RAT and RAD interventions with CwASD, in general, are different, many of the RAT’s advances and positive findings can be directly or indirectly applied in the development of applications for autism screening using a robotic-assistive approach. For example, the already clear need to use well-standardized clinical protocols, the use of good and rigorous experimental designs to generate clinical evidence [5], and the findings regarding the acceptability of robot’s physical features and automated behaviors [12], [13] are all applicable to improve the screening processes and identification of risk factors using the paradigm of children’s interaction with a robot-mediator. Thus, RAT researches have shown relevant behavioral domains where the benefits of CRI are usefully and promising, such as Joint Attention (JA), imitation task, and free play [4], [14], [15]. The child’s response towards a robotic platform in these domains also can be used as diagnostic inputs [6].

Thus, the aim of this work is provides a comprehensive Robot-Assisted Intervention for CwASD, showing the conditions in which a robot-based approach can be useful to assess autism risk factors for an autism diagnosis purpose. More specifically, in this work, the behavior of CwASD is compared with that of children in a control group during a human/robot mediation while JA behaviors are elicited and analyzed. The primary objective of this study is to identify behavior patterns associated with JA impairments and autism risk factors during a single exposed robot interactive session. Second, the advantage of using the CRI to emerge patterns’ differences between the two groups of children was also investigated.

This study uses important findings from previous works in RAT domain. For this reason, in this work, several important methodological strengths are proposed, such as a well-established protocol for JA assessment based on Discrete Trial Teaching (DTT), a group-based experimental design for two kind of conditions and the improvement of the technological framework based on a multimodal environment for robot-mediated intervention (MERI) developed in our previous study [16].

Section snippets

Robot-based intervention for autism screening

The justification of Tapus et al. [17] for using robots as autism screening tools continues to be valid. A robotic system is understood as an interactive platform composed of many subsystems with the ability to obtain information about the children–robot interaction passively as well as actively and to provide systemically social cues designed to elicit particular social responses [2]. The use of a robotic platform allows creating systematic social prompts to stimulate children and generate

New Multimodal Environment for Robot-based intervention (MERI)

All technical components related to the new robot-assisted diagnosis scheme are coordinated and synchronized using an improved version of the MERI system proposed in our previous work [16]. The new approach of MERI aims to wrapper multiple stages of an autism diagnostic process, such as protocol planning and execution, data recording and intervention analysis, with a single user interface. The updates of the system are explained below.

The current MERI version remains as a ROS-based1

Procedures

The MERI system was adapted and installed in the music-therapy room of the Colombian Rehabilitation Clinic, Howard Gardner (HG). The HG clinic specializes in therapies for children with special needs and with neurodiversity conditions. In this therapeutic institute, four kinds of personalized interventions are supplied: (i) occupational therapy, (ii) speech and language therapy, (iii) physical therapy, and (iv) psychology, which are given daily to each of their patients (young individuals from

Results

Fig. 4 shows the MERI Interface designed and implemented to execute all necessary actions in the RAD intervention. The video processing algorithm was validated in our previous work [16] and in the scenarios of this work achieved a child’s face tracking on 87% of frames in the 240 s session interval selected for analysis. The mean processing rate in frame per second (FPS) in all sessions was superior to 24 fps. In addition, the head orientation and VFOA were estimated in all video, thanks to the

Discussion

From an autism diagnostic perspective, the results are promising, given that they allow exhibiting a series of useful findings to identify autism risk factors associated mainly with CwASD ability to manifest JA behaviors when a robot elicits them. It is clear that the ASD manifestation signs are vast, and our goal was not to find large patterns. However, from the perspective of the diagnosis, this application seems to show some common signs in children that become more evident when interacting

Conclusion

In this study, a robot-assisted diagnosis technique based on a MERI system was proposed and tested. Through the MERI user interface, JA stimuli were provided, recorded, and analyzed successfully using a well-structured protocol. While social robotic applications are considered a potential tool to elicit differential behaviors, few studies have shown a technological tool to cover all robot intervention stages using experimental designs relevant to core JA challenges.

In conclusion, children with

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We are thankful to the participants and their families for participated in this study. Also, we thank the Howard Gardner Clinic, Colombia for the support. finally, the first author thanks Coordenãço de Aperfeiçoamento de Pessoal de Nivel Superior - Brasil (CAPES) - for the scholarship (Finance Code 001 ) and the Royal Academy of Engineering, United Kingdom - CASTOR Project: CompliAnt SofT Robotics (grant IAPP1\100126).

Andrés A. Ramírez-Duque received his bachelor’s degree in Mechatronics Engineering from the Universidad Nacional de Colombia, Bogotá, Colombia, in 2009, and his Industrial Automation Master degree from the Universidad Nacional de Colombia, Bogotá, Colombia, in 2011. He received his Ph.D. in Electrical Engineering from the Federal University of Espírito Santo, Vitória, Brazil, in 2019. He won a Google Latin America Research Award 2017. He is currently professor with the El Bosque University

References (43)

  • ScassellatiB. et al.

    Robots for use in Autism research

    Annu. Rev. Biomed. Eng.

    (2012)
  • SteinerA.M. et al.

    Practitioner’s guide to assessment of Autism spectrum disorders in Infants and Toddlers

    J. Autism Dev. Disord.

    (2012)
  • WigginsL.D. et al.

    Using standardized Diagnostic instruments to classify children with Autism in the study to explore early development

    J. Autism Dev. Disord.

    (2015)
  • Baron-CohenS. et al.

    Can autism be detected at 18 Months?: The Needle, the Haystack, and the CHAT

    Br. J. Psychiatry

    (1992)
  • LecavalierL.

    An evaluation of the Gilliam Autism rating scale

    J. Autism Dev. Disord.

    (2005)
  • SimutR. et al.

    Children with Autism spectrum Disorders make a fruit salad with Probo, the social robot: An interaction study

    J. Autism Dev. Disord.

    (2016)
  • CabibihanJ.J. et al.

    Why robots? A survey on the roles and benefits of social robots in the Therapy of children with Autism

    Int. J. Soc. Robot.

    (2013)
  • HuijnenC.A.G.J. et al.

    Mapping robots to Therapy and educational objectives for Children with Autism Spectrum disorder

    J. Autism Dev. Disord.

    (2016)
  • HuijnenC.A.G.J. et al.

    Matching robot KASPAR to Autism spectrum disorder (ASD) Therapy and educational goals

    Int. J. Soc. Robot.

    (2016)
  • Ramírez-DuqueA.A. et al.

    Robot-assisted Autism Spectrum disorder diagnostic based on artificial reasoning

    J. Intell. Robot. Syst.

    (2019)
  • TapusA. et al.

    Socially assistive robotics [Grand Challenges of Robotics]

    IEEE Robot. Autom. Mag.

    (2007)
  • Cited by (0)

    Andrés A. Ramírez-Duque received his bachelor’s degree in Mechatronics Engineering from the Universidad Nacional de Colombia, Bogotá, Colombia, in 2009, and his Industrial Automation Master degree from the Universidad Nacional de Colombia, Bogotá, Colombia, in 2011. He received his Ph.D. in Electrical Engineering from the Federal University of Espírito Santo, Vitória, Brazil, in 2019. He won a Google Latin America Research Award 2017. He is currently professor with the El Bosque University (Bogotá, Colombia). His current research interests include Child–Robot interaction, cloud parallel computing, high performance computing, smart environments and serious games applied to Children with development and learning impairments.

    Teodiano Bastos received the B.S. degree in electrical engineering from the Federal University of Espirito Santo, Vitória, Brazil, in 1987, and the Ph.D. degree in physical science (Electricity and Electronics) from the Universidad Complutense de Madrid, Spain, in 1994. He is currently a full professor with the Federal University of Espirito Santo. His current research interests are biomedical signal processing, rehabilitation robotics and assistive technology.

    Marcela Munera received her Ph.D. in Mechanics and Biomechanics from Université de Reims Champagne Ardenne thanks to a FEDER, Region Champagne Ardenne–Doctoral Grant. She graduated as a Bioengineer from Universidad de Antioquia and from the Ecole Nationale de Metz with a Masters in Mechanics and Materials. During her Ph.D. and after, as a Lecturer at Université de Reims Champagne Ardenne (France), she worked in industrial research projects in biomechanical assessment related to sports performance and injury prevention. In sports, her research was mainly experimental and focused on vibratory and dynamic response, human effects of shock and vibration and wearable sensors. Currently, she is an assistant professor of Biomedical engineering at the Escuela Colombiana de Ingenieria Julio Garavito, where she contributes to projects in rehabilitation, particularly in the objective assessment of robotic devices and systems in different scenarios, and the assessment of the human response at the Center for Biomechatronics. She is a Member of the European Society of Biomechanics (since 2014), and act as reviewer for several engineering and multidisciplinary journals. Her research interests are focused in biomechanics, movement analysis and assessment in rehabilitation and sports. Other research interests involve understanding the human factors in experimental interventions of robotics in rehabilitation.

    Carlos A. Cifuentes is a Professor with the Department of Biomedical Engineering and Head of the Center for Biomechatronics at the Escuela Colombiana de Ingenieria Julio Gravito (ECIJG-Colombia). He has been a Visiting Professor at the Universidade Federal do Espirito Santo, University of Cauca, Plymouth University and the EPF Graduate School of Engineering. Prior to that, he was a postdoc at Universidade Federal do Espírito Santo (UFES-Brazil). He is broadly interested in human–robot interaction and rehabilitation robotics in the context of developing countries. He was born in Bogota, Colombia. In 2004, he received a BSc degree in Electronic Engineering from the ECIJG. He received his Specialization in Project Management in 2006 at ECIJG. In 2011, he obtained his M.Sc. degree in Biomedical Engineering from the Universidad Nacional de Entre Rios, Argentina. In 2012, he joined the Robotics and Industrial Automation Group at UFES to pursue his Ph.D. degree. He developed a part of his thesis at the Automation Institute, (UNSJ-Argentina) and at the Neural and Cognitive Engineering group, CAR, (UPM-CSIC-Spain). His Ph.D. thesis (2015), for which he received the Honorable Mention Award CAPES as one of the best theses in 2016 in Brazil, focused on develop a multimodal human–robot interface that provides a means of testing and validating control strategies for robotic walkers for assisting human mobility and gait rehabilitation. In 2017 his work was lauded as one of “five history-changing ideas in Latin America” by the History Channel.

    Anselmo Frizera-Neto received the B.S. degree (2006) in Electrical Engineering from the Federal University of Espirito Santo (UFES), Vitória, Brazil and the Ph.D. in Electronics (2010) from the Universidad de Alcalá, Spain. From 2006 to 2010, he was a Researcher with the Bioengineering Group, Spanish National Research Council (CSIC). He is currently a Professor in the Department of Electrical Engineering, UFES. His research interests include rehabilitation robotics, human–machine interaction, and movement analysis.

    View full text