Interactive Mirroring Games wIth sOcial Robot (IOGIOCO) and Robotic System Adapted Into a Clinical Scale (RISCALE)

Background: The importance of early intervention in Autism Spectrum Disorders (ASD) has been widely demonstrated, and developmental trajectories in ASD highlight the importance of nonverbal communication, such as intransitive gesture production, as a possible positive prognostic factor for language development. The use of technological tools in the therapy of individuals with ASD has also become increasingly important due to their increased engagement and responsiveness to technological objects, such as robots.

We developed a training protocol using the humanoid robot NAO, called IOGIOCO (Interactive mirroring Games wIth sOCial rObot), based on the use of intransitive gestures embedded in naturalistic dialogues, stimulating a triadic interaction between child, robot and therapist. The training is divided into six levels; the first 2 levels were called "familiarization levels," and the other 4 were "training levels". The technological setup includes different complexity levels, from mirroring tasks to building spontaneous interactions. We previously tested the protocol on 10 preschool children with ASD (aged 2-6 years) in a pilot study founding promising results (Annunziata et al,2024). We therefore developed a Randomized Controlled Trial to test the efficacy of this intervention.

The primary objective of the RISCALE study is to evaluate the effectiveness of NAO training in improving socio-communicative skills and acquiring communicative gestures using the ECSP-I scale as the primary outcome measure.

A second objective stems from the technological experience acquired in these two years on the development of methods for the quantitative recording of social interaction (child-operator, child-NAO gaze contact, joint attention...). 2 scenarios have been identified in the scale that can be achieved both with the human operator and through the action of NAO:

• the response to the adult's gesture of indication;
• the differential response - Joint Attention and/or Behavior Regulation - to the activation of a mechanical toy).

The two scenarios will be videotaped with Kinect equipment and measurable data will be acquired for quantitative analysis. The procedure will be validated on a sample of typically developing children (about 20); adapted and validated the instrument on the sample of typically developing children, then the administration and then the validation of the setup obtained on subjects with Autism Spectrum Disorder will take place.

The aim is to evaluate whether and how the modes of visual exploration and joint attention change and/or differ pre- and post-training, and the differences that may be evident in the modes of interaction between child and operator and between child and robot.

In time, therefore, the following will be provided:

1. to set, adapt and subsequently validate the video recording setting with the identification of the quantitative parameters detectable through technological tools (ECSP-I_rob) on a sample of typically developing children aged between 18 and 24 months
2. to set, adapt and subsequently validate the video recording setting with the measurement of the quantitative parameters detectable through technological tools (ECSP-I_rob) on a sample of children with Autism Spectrum Disorder under the age of 72 months
3. start training with NAO on a group of children with ASD aged 24-72 months with pre- and post-treatment ECSP-I administration.

Objectives

1. To evaluate the effectiveness of the robot-based intervention on social-communicative skills and in the acquisition of intransitive gestures used for communicative purposes using the ECSP-I and the MacArthur questionnaire as an outcome measure. The rationale remains the one already detailed in the previous phase, namely that ASD children tend to have greater difficulty in paying attention to the variability and multiplicity of signals characterizing human social interaction (e.g. mimicry, gestures...), an aspect that could lead to a possible lower interest in interaction with an individual than in a technological object towards which they would seem more attentive, responsive and quick in response
2. identify and describe measurable parameters of "social" interaction within the rehabilitation session with NAO and operator and verify their evolution and differences at the beginning and end of treatment and between the different social partners.

Translationality The present study will allow to verify the feasibility and usability of the experimental evaluation protocol based on the use of the social robot NAO.

If its validity is demonstrated, this method could help to make habilitative interventions in children with ASD more motivating and effective, favoring the generalization of the skills acquired in the context of daily life.

Methodology Study design Multicenter Experimental Randomized Open-Label Controlled Study

Type of study Low-risk experimental study

Case studies Phase 1: 20 typically developing children aged 18-24 months for ECSP-I_rob data acquisition (videotaped ECSP-I administration) Phase 2: 10 ASD children - videotaped ECSP-I administration Phase 3: 20 children diagnosed with ASD aged between 24 and 72 months, already in charge for rehabilitation treatment, who will subsequently be divided into two groups for treatment.

Sample Recruitment The 20 typically developing children will be recruited for adaptation and validation of the ECSP-I scale to the NAO interaction system on a voluntary basis at the NEARLab of the Politecnico di Miano and will be evaluated by a properly trained TNPEE of Fondazione Don Gnocchi

Children with ASD will be recruited from among the subjects who belong to the outpatient clinics or outpatient rehabilitation treatments at IRCCS Fondazione Don Gnocchi. The sample will be recruited sequentially among children in charge for rehabilitation treatment, who have a diagnosis of Autism Spectrum Disorder confirmed with the main gold-standard tools.

Sample size The calculation of the sample size is based on the primary objective of the study, i.e. the evaluation with the ECSP-I scale of the treated group and the control group.

Using the Mann-Whitney Test for independent samples to compare the two groups, with a standard deviation estimate of 0.56 and a study design with a 1:1 ratio between the treated group and the control group, it is calculated that a sample of 10 subjects from each group is required to highlight an expected difference of 0.8 points on the scale with a study power of 80% and a level of significance of 5%.

Statistical analysis A sequential design will be adopted for the comparison between the two groups. The adoption of this design will allow to obtain a sample of 20 participants for pre- and post-training comparison and to offer the training provided for the experimental group (group A) also to the initially control group (group B) without substantial burdens for the standard therapy procedure; it will also allow to verify the change due exclusively to the passage of time and standard therapy (T0 and T1 comparison for group B) and to carry out a follow-up (T1 and T2 comparison for group A). We also expect greater adherence to the trial, as parents would be enticed to participate by inclusion in an additional pathway to standard therapy (and not simply participation as a control group).

Given the expected number of participants, and the scores of the ECSP-I scale that are expressed in ordered categories (Optimal Levels), the comparison between the measurements provided for by the sequential design (T0, T1 and T2) will be carried out with non-parametric statistical tests:

• Wilcoxon Exact Test, for longitudinal comparison of each group.
• test for independent samples in the exact form test for independent samples in the exact form (Mann-Whitney Exact Test), for the comparison between the experimental group and the control group.
• The 2 groups will also be compared through a multivariate analysis, conducted by applying the ANCOVA model adjusted for months of age and baseline scale values to correct for potential imbalances due to the low sample size.

Procedures

Phase 1: ADMINISTRATION OF ECSP-I SCALE on a typically developing sample The 20 typically developing children will be recruited for the adaptation and validation of the two scenes of the ECSP-I scale to the NAO interaction system on a voluntary basis at the NEARLab of the Politecnico di Miano and the scale will be administered by a properly trained TNPEE of Fondazione Don Gnocchi

Phase 2: ADMINISTRATION OF THE ECSP-I SCALE and robot-adapted scenes to a sample of children with ASD The ECSP-I scale and scenes adapted to the social robot will be administered to a sample of children with Autism Spectrum Disorder in order to gain experience and collect data about the administration of this scale to a clinical sample of 10 children.

The administration will be videotaped and encoded with the supervision of the developers.

Phase 3: Recruitment and randomization of the sample of children with ASD. 20 children with Autism Spectrum Disorder will be recruited, subjected to T0 assessment.

Phase 4: The sample will then be divided into two groups:

• a group A and group B, which will carry out training with the social robot (see Appendix A for the training protocol) at 2 different times.

Phase 5: All the recruited sample will be subjected, at different times, to rehabilitation training with NAO following the protocol in the Appendix. The training will last 12 weeks, with one session per week, lasting 20-30 minutes.

Phase 6: Re-evaluation to T1 and T2 according to the treatment scheme (see Table 3).

Evaluation tools

• Griffiths Mental Development Scales III ed (GMDS-III) [30]
• Adaptive Behavior Assessment System II ed (ABAS II) [31]
• Autism Diagnostic Observation Shedule (ADOS 2) [32, 33]
• Autism Diagnostic Interwiev Revised (ADI-R) [34]
• McArthur-Bates Questionnaire, First Vocabulary of the Child (PVB) Gestures and Words: total number of gestures [35]
• Early Social Communication Scale (ECSP-I) [27]
• Adapted scenes of the ECSP-I with NAO Social Robot (ECSP-I_rob) (quantitative measurement of interaction data)

Criticality The main critical issue concerns the recruitment of subjects and their availability during the study period and the continuity of the intervention. For this reason, an extended recruitment period has been hypothesized, which will cover a large part of the entire duration period of the study.

Expected results

• Verification of the sensitivity of the ECSP-I in grasping the improvement of socio-communicative skills, in particular in the use of gestures.
• Improvement in the use of gestures learned during training, in terms of: - Correctly learned gestures used in the therapeutic setting in known scenarios; - Learned gestures correctly used in the therapeutic setting in new scenarios (first step of generalization);
• Increase in the number of gestures used in daily life (second phase of generalization; Mc Arthur questionnaire);
• Measurement of social interaction parameters detectable with technological tools.