Utilizing Large Language Models to Augment Family Conversations in the Intensive Care Unit
试验速览
- 阶段
- 不适用
- 状态
- 招募中
- 发起方
- 入组人数
- 40
- 试验地点
- 1
- 主要终点
- Documentation quality of family conversations
研究概览
简要总结
This study looks at how artificial intelligence (AI), like generative pre-trained transformer (GPT-4), can help doctors in the intensive care unit (ICU) save time and improve communication with families. Right now, doctors spend a lot of time writing notes after family conversations, which takes time away from patient care. The investigators are testing whether AI can create accurate and easy-to-understand summaries of these conversations, making it quicker for doctors to document and clearer for families to understand. ICU doctors and adult family members of patients will take part in this study, with their full consent. The goal is to see if this new technology can make life easier for doctors while helping families better understand medical information.
研究设计
- 研究类型
- Observational
- 观察模型
- Cohort
- 时间视角
- Prospective
入排标准
- 年龄范围
- 18 Years 至 —（Adult, Older Adult）
- 性别
- All
- 接受健康志愿者
- 否
入选标准
- •Informed-consent is signed by all participating family members during the conversation.
- •Informed-consent is signed by the ICU healthcare professionals.
- •The patient of the family must be admitted to the adult ICU.
- •All participating family members must be 18 years or older.
- •The family conversation must be in Dutch.
排除标准
- •Trained to interpret medical jargon and/or intensive care terminology specifically
- •Difficulty reading and/or writing in the Dutch language
结局指标
主要结局
Documentation quality of family conversations
时间窗: Within 30 days after the family conversation.
The primary outcome measure is the documentation quality of family conversations, evaluated using a modified Physician Documentation Quality Instrument-9. Originally validated for progress notes and discharge summaries, Physician Documentation Quality Instrument-9 scores notes on nine attributes (accurate, thorough, useful, organized, comprehensible, succinct, synthesized, consistent, up-to-date) using a 1-5 scale (1 = "not at all," 5 = "extremely"), for a total ranging from 9 to 45. To adapt it for AI-generated text, the investigators removed the "up-to-date" domain (focused on whether the note includes the latest test results/recommendations) and added two new attributes: freedom from hallucinations (unfounded information) and freedom from bias (discriminatory data, algorithms, or heuristics). These additions address potential pitfalls in large language model outputs, such as introducing incorrect content or skewed results.
次要结局
- Time spent on documentation by ICU clinicians(Within one day after the family conversation.)
- Family satisfaction(Within 30 days after the family conversation.)
研究者
Davy van de Sande
Coordinating Investigator
Erasmus Medical Center