预测阿片类药物使用障碍药物的治疗保留:使用NLP和LLM衍生的临床特征的机器学习方法
Fateme Nateghi Haredasht1, Ivan Lopez2,3, Steven Tate4
1Stanford Center for Biomedical Informatics Research, Stanford, CA 94305, United States.
Journal of the American Medical Informatics Association : JAMIA
|September 22, 2025
概括
大型语言模型 (LLMs) 通过分析临床笔记来提高预测布诺芬-纳洛 (BUP-NAL) 治疗保留率. 整合LLM功能可以提高准确性,并识别关键的心理社会风险,以改善患者护理.
科学领域:
- 人工智能在医学中的应用
- 临床信息学 临床信息学
- 药物使用障碍治疗 药物使用障碍治疗
背景情况:
- 预测阿片类药物使用障碍的治疗保留对于患者的治疗结果至关重要.
- 布普伦诺芬-纳洛 (BUP-NAL) 治疗是关键的治疗方法,但保留预测需要改进.
- 非结构化临床笔记包含有价值的信息,这些信息并不总是包含在结构化EHR数据中.
研究的目的:
- 为了提高BUP-NAL治疗的6个月保留预测.
- 将从应用到临床笔记的大型语言模型 (LLM) 中获得的特征集成.
- 为改善接受BUP-NAL治疗的患者个性化风险分层.
主要方法:
- 使用非识别的EHR数据用于模型开发和验证.
- 从自由文本笔记中提取了13个临床和心理社会特征,使用基于LLM的管道.
- 训练和评估各种分类和生存模型,包括XGBoost和随机生存森林,使用ROC-AUC和C指数.
主要成果:
- 在所有测试的模型架构中,LLM衍生的功能改善了预测性能.
- XGBoost获得了最高的分类性能 (ROC-AUC=0.65),随机生存森林/生存XGBoost显示了最高的C指数 (≈0.65).
- 长期疼痛,肝脏疾病和严重抑郁症等LLM提取的特征被确定为重要的预测因素.
结论:
- 通过NLP和LLM辅助的特征提取可以提高预测的准确性和可解释性.
- 这些方法揭示了仅仅通过结构化EHR数据错过的心理社会风险.
- 将结构化数据与LLM功能相结合,可以适度地改善AI驱动护理的BUP-NAL保留预测.
更多相关视频
06:22Machine Learning-Based Cough Tone Classification: Diagnostic Exploration of Chronic Obstructive Pulmonary Disease and Respiratory Tract Infections
Published on: September 19, 2025
435
08:53Integrating Computerized Linguistic and Social Network Analyses to Capture Addiction Recovery Capital in an Online Community
Published on: May 31, 2019
5.5K
相关概念视频
01:15Drug Abuse and Addiction: Pharmacological Phenomena
1.1K
Drug dependence, abuse, and addiction are complex phenomena that can precipitate various abnormal states. Physical dependence refers to a state of pharmacological adaptation to a drug. This adaptation often results in tolerance—a reduced response to the drug after repeated administrations. When the drug use is abruptly stopped, withdrawal symptoms occur due to the body's need to readjust from the pharmacologically induced imbalance. However, tolerance and withdrawal symptoms do not...
1.1K
01:25Analgesia and Pain Management
1.5K
Pain is critical to various clinical pathologies, provoking an urgent need for effective management. Pain, whether acute or chronic, is a complex neurochemical process. Its alleviation depends on the type, with nonopioid analgesics effective for mild to moderate pain, such as musculoskeletal or inflammatory pain, while neuropathic pain responds best to anticonvulsants, tricyclic antidepressants, or serotonin/norepinephrine reuptake inhibitors. For severe acute or chronic pain, opioids may be...
1.5K