Impact of a Social Robot on Hospitalized Children, Caregivers, and Health Care Staff: Exploratory Observational Study

Background Hospitalization poses significant psychosocial challenges for children, including anxiety, emotional distress, and disruption of daily routines. While animal-assisted therapy has shown benefits, practical barriers such as infection control concerns and inconsistent availability limit its implementation in pediatric wards. Social robots have emerged as a promising alternative, but evidence on their psychosocial impact in pediatric settings remains limited. LOVOT is a companion robot designed for emotional engagement through warmth, softness, and nonverbal interaction; however, no studies have examined its effects in pediatric health care. Objective This exploratory study aimed to assess the perceived impact of LOVOT on hospitalized children, their caregivers, and health care staff in a pediatric ward. Methods This prospective observational study was conducted in a pediatric ward at Hyogo Prefectural Amagasaki General Medical Center, Japan, from May 2024 to March 2025. Two LOVOT units were permanently installed in the playroom for free interaction. Caregivers (n=110) and health care staff (n=32) completed postintroduction questionnaires with retrospective ratings of perceived change using 5-point Likert scales (1=“much worse” to 5=“much better,” with 3=“no change” as the baseline). Free-text responses (n=91) were analyzed using automated emotion classification. One-sample 2-tailed t tests or Wilcoxon signed-rank tests compared ratings against the baseline, with false discovery rate (FDR) correction for multiple comparisons. Results All 10 caregiver-rated and 11 of 12 staff-rated outcomes showed significant perceived improvement (all caregiver items FDR P <.001; staff items FDR P ≤.02; the workload item was not significant, P =.37). Mean ratings ranged from 3.28 to 4.62 on the 5-point scale (baseline: 3=“no change”). The largest perceived improvements were reported for children’s enjoyment of hospital life (mean 4.47, SD 0.62), adaptation to hospital life (mean 4.10, SD 0.70 to mean 4.28, SD 0.63), and stress and anxiety reduction (mean 4.17, SD 0.76 to mean 4.19, SD 0.64). One-sample effect sizes ranged from medium to very large (Cohen d =0.53-2.67); however, because these were calculated against a fixed neutral baseline rather than a control group, they should be interpreted cautiously. Ratings from caregivers and staff showed strong agreement on 4 of 5 matched items. Notably, no detectable increase in staff workload was observed (mean 3.09, SD 0.53; P =.37), supporting implementation feasibility. Qualitative analysis revealed predominantly positive responses, with joy as the dominant emotion (mean probability 88.7%, 95% CI 86.8%-90.6%). Conclusions This study provides preliminary evidence that a companion robot (LOVOT) may positively impact hospitalized children, caregivers, and health care staff across multiple psychosocial outcomes without a detectable increase in staff workload. The low response rate and reliance on retrospective self-report limit generalizability and causal inference; however, convergent findings across stakeholder perspectives justify further investigation through controlled trials. Companion robots represent a feasible, low-burden adjunct to psychosocial care in pediatric settings.