The Impact of Conversational Approaches in Pediatric Robot-Mediated Interventions on Children's Responses in Hospital Settings: A Comparative Study of Rule-Based vs. LLM-Based Systems
试验速览
- 阶段
- 不适用
- 状态
- 尚未招募
- 入组人数
- 60
- 试验地点
- 1
- 主要终点
- Child Engagement Level
研究概览
简要总结
This study examines how different robot dialogue systems (rule-based vs. large language model-based) and content types (emotional support vs. safety education) affect pediatric patients' responses during hospital-based robot-mediated interventions.
Approximately 60 pediatric patients aged 2-9 years will be randomly assigned to interact with a social robot (LIKU) using either rule-based or LLM-based dialogue. Each child will participate in two activity sessions (emotional content and safety content) in randomized order.
Primary outcomes include child engagement, emotional responses, robot perception, and activity preferences, assessed through standardized questionnaires (UEQ, Godspeed), child interviews, and behavioral observations. Additionally, 5 experts will evaluate content appropriateness and safety.
This pilot study aims to provide foundational data for developing personalized pediatric robot programs in hospital settings, optimizing both dialogue approaches and content design based on individual child characteristics.
研究设计
- 研究类型
- Interventional
- 分配方式
- Randomized
- 干预模型
- Factorial
- 主要目的
- Supportive Care
- 盲法
- Single (Outcomes Assessor)
入排标准
- 年龄范围
- 2 Years 至 9 Years（Child）
- 性别
- All
- 接受健康志愿者
- 是
入选标准
- •Pediatric patients aged 2-9 years
- •Currently admitted to or visiting Samsung Medical Center pediatric ward or outpatient clinic
- •Capable of cognitive communication as determined by experienced medical staff
- •Written informed consent from parent/legal guardian obtained
- •Verbal assent from child obtained (developmentally appropriate)
- •Not in acute emergency situation
- •Able to engage in stable interaction for at least 5 minutes in hospital environment
排除标准
- •Diagnosis of moderate to severe developmental delay (language or cognitive) documented in medical records
- •Currently in isolation treatment (e.g., due to infectious disease)
- •Experiencing acute pain, fever, or medical procedures that make study participation difficult
- •Parent/guardian or child declined participation or withdrew consent
研究组 & 干预措施
Rule-Based Robot Dialogue
Participants interact with LIKU robot using pre-programmed, rule-based dialogue responses. Each participant completes two activity sessions (emotional support content and safety education content) in randomized order.
干预措施: Robot-Mediated Emotional Support Activity (Behavioral)
Rule-Based Robot Dialogue
Participants interact with LIKU robot using pre-programmed, rule-based dialogue responses. Each participant completes two activity sessions (emotional support content and safety education content) in randomized order.
干预措施: Robot-Mediated Safety Education Activity (Behavioral)
LLM-Based Robot Dialogue
Participants interact with LIKU robot using large language model-generated dialogue responses. Each participant completes two activity sessions (emotional support content and safety education content) in randomized order.
干预措施: Robot-Mediated Emotional Support Activity (Behavioral)
LLM-Based Robot Dialogue
Participants interact with LIKU robot using large language model-generated dialogue responses. Each participant completes two activity sessions (emotional support content and safety education content) in randomized order.
干预措施: Robot-Mediated Safety Education Activity (Behavioral)
结局指标
主要结局
Child Engagement Level
时间窗: Immediately following each intervention session (assessed within 1 hour post-session)
Behavioral observation checklist assessing attention to robot, participation in activities, and task involvement. Scored by trained observers using standardized coding scheme based on video recordings.
User Experience
时间窗: Immediately following each intervention session (approximately 5 minutes post-session)
User Experience Questionnaire (UEQ) - standardized 7-point semantic differential scale measuring six dimensions: attractiveness, perspicuity, efficiency, dependability, stimulation, and novelty of robot interaction.
Robot Perception
时间窗: Immediately following each intervention session (approximately 5 minutes post-session)
Godspeed Questionnaire Series - standardized scale measuring five dimensions of robot perception: anthropomorphism, animacy, likeability, perceived intelligence, and perceived safety.
次要结局
- Content Appropriateness(Within 2 weeks of session completion via video review)
- Activity Preference(Immediately following completion of both intervention sessions (approximately 5 minutes))
- Content Comprehension(Immediately following each intervention session (approximately 3 minutes))
- Robot Relationship Quality(Immediately following intervention sessions (approximately 5 minutes))