Feasibility of a Robot-Guided Active Play Intervention for Children With Cerebral Palsy (REACT)

Participants will complete one in-person, NAO-delivered, movement-based narrative active-play session. The intervention includes parallel ambulatory (GMFCS I-III) and non-ambulatory (GMFCS IV) versions, with both versions sharing a common narrative structure and movement library.

A humanoid social robot (NAO, SoftBank Robotics) guides the participant through a single, story-based active-play session in which physical movements are embedded within a narrative. The intervention is designed to promote engagement and active participation rather than conventional prescriptive exercise delivery. Parallel ambulatory (GMFCS I-III) and non-ambulatory (GMFCS IV) versions are used, with both sharing a common narrative structure and movement library.

• REACT

Number of child/adolescent participants enrolled relative to the prespecified recruitment target. Recruitment will be summarized descriptively. Progression will be interpreted as: target achieved = proceed; below target but recruitment achievable with an adjusted strategy or timeline = proceed with modification; below target with no feasible recruitment pathway = reconsider.

Percentage of eligible child-caregiver dyads invited to participate who provide caregiver consent and child/adolescent assent and enroll in the study. Prespecified progression thresholds are ≥80% = proceed; 60-79% = proceed with modification; <60% = reconsider.

Percentage of enrolled child/adolescent participants who complete the planned NAO-guided active-play intervention session. Prespecified progression thresholds are ≥80% = proceed; 60-79% = proceed with modification; <60% = reconsider.

Fidelity to the prespecified NAO-guided intervention protocol will be assessed using the approved intervention fidelity checklist and summarized as a percentage. Prespecified progression thresholds are ≥80% = proceed; 60-79% = proceed with modification; <60% = reconsider.

Number and nature of protocol deviations occurring during study participation, recorded using the approved protocol deviation and adverse event log and summarized descriptively.

Number and nature of adverse events occurring during study participation, recorded using the approved protocol deviation and adverse event log and summarized descriptively.

Child/adolescent engagement and intention to repeat the activity will be assessed using the age-appropriate REACT Post-Narrative Intervention Likert Scale (child version, ages 6-13; adolescent version, ages 14-18). Each version contains five items assessing enjoyment, desire to continue the activity, ease of performing the movements, experience interacting with NAO, and willingness to participate in a similar session again. Each item is rated from 1 to 5, with higher scores indicating a more positive response on that item. Responses will be summarized descriptively at the item level; no composite or total score will be calculated.

Caregiver acceptability and willingness will be assessed using the purpose-built REACT Caregiver Acceptability and Willingness Survey. The survey contains 10 Likert items assessing perceived child enjoyment, engagement, comfort and safety, overall experience, willingness to participate again, willingness to participate in a longer program, comfort with use in therapy or at home, and willingness to recommend the session to other families. Each item is rated from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating greater agreement and a more positive response on that item. Responses will be summarized descriptively at the item level; no composite or total score will be calculated.

Child/adolescent perceptions of intervention length and comfort while wearing the arm-strap heart-rate monitor and wrist-worn accelerometer will be assessed using purpose-built questions and prompts in the REACT Facilitator Interview Guide. Participants will be asked whether the activity felt too long, too short, or appropriate in length and whether the study sensors caused discomfort or were noticeable during participation. Responses will be analyzed qualitatively; no numerical scale or composite score will be calculated.

A semi-structured dyadic interview with the child/adolescent and caregiver will explore their experience of the NAO-guided active-play intervention, including perceived acceptability, what worked or did not work, barriers and facilitators to participation, and recommended refinements to the intervention and study procedures.

Child/adolescent perceived exertion will be assessed using the cerebral palsy-specific OMNI Rating of Perceived Exertion scale, with ambulatory and non-ambulatory pictorial versions used as appropriate. The scale ranges from 0 ("not tired at all") to 10 ("very, very tired"), with higher scores indicating greater perceived exertion. Higher scores do not inherently represent a better or worse outcome.

Heart rate will be recorded using an arm-strap heart rate monitor during a seated resting baseline and the NAO-guided active-play intervention. Heart rate data will be summarized descriptively to characterize physiological response during participation.

Movement data will be collected using a wrist-worn accelerometer and summarized descriptively to characterize movement and physical activity during the NAO-guided active-play intervention.

Wear time for the arm-strap heart-rate monitor and wrist-worn accelerometer will be recorded in minutes from initial sensor placement to removal. Wear time will be summarized descriptively for each sensor separately; no combined wear-time score will be calculated.

The number of heart-rate monitor and accelerometer removal events occurring during the planned sensor-wear period will be recorded and summarized descriptively as counts. Events will be reported by sensor type.

Data completeness will be assessed separately for the arm-strap heart-rate monitor and wrist-worn accelerometer as the percentage of the planned recording period for which sensor data are available. Higher percentages indicate greater data completeness. Heart-rate and accelerometer completeness will be reported separately; no combined score will be calculated.

Behavioural engagement during the NAO-guided intervention will be assessed from video recordings using a study-specific structured coding protocol developed and piloted by two independent coders. Behavioural engagement codes will be summarized descriptively; no standardized scale or composite score will be calculated. Inter-rater reliability will be established on a minimum of 20% of recordings, with a target Cohen's κ ≥0.70 before full coding.

Movement quality during the NAO-guided intervention will be assessed from video recordings using a study-specific structured coding protocol developed and piloted by two independent coders. Movement-quality codes will be summarized descriptively; no standardized scale or composite score will be calculated. Inter-rater reliability will be established on a minimum of 20% of recordings, with a target Cohen's κ ≥0.70 before full coding.

Critical incidents occurring during the NAO-guided intervention will be identified from video recordings using a study-specific structured coding protocol developed and piloted by two independent coders. Critical-incident codes will be summarized descriptively; no standardized scale or composite score will be calculated. Inter-rater reliability will be established on a minimum of 20% of recordings, with a target Cohen's κ ≥0.70 before full coding.