PH-LLM:公共卫生信息监控的大型语言模型
Xinyu Zhou1,2, Jiaqi Zhou1,2, Chiyu Wang3
1Division of Biostatistics and Informatics, Department of Preventive Medicine, Northwestern University, Chicago, IL 60611, USA.
medRxiv : the preprint server for health sciences
|February 24, 2025
概括
公共卫生机构现在可以使用PH-LLM (公共卫生信息监控大型语言模型) 实时监控社交媒体. 这种人工智能工具提供了先进的多语言功能,用于更好的公共卫生响应和政策制定.
科学领域:
- 人工智能的人工智能
- 公共卫生信息学 公共卫生信息学
- 计算语言学 计算语言学
背景情况:
- 公共卫生干预需要公众的支持,通常受到社交媒体话语的影响.
- 实时社交媒体监控对于及时的公共卫生响应至关重要,特别是在紧急情况下.
- 现有的监测方法缺乏在线有效分析公众健康情绪的速度和范围.
研究的目的:
- 开发一套新的大型语言模型 (LLM) 套件,用于实时公共卫生信息监控.
- 创建一个能够在社交媒体上监控公共卫生问题的多语言AI系统.
- 为了解决为公共卫生政策调整提供及时数据分析的差距.
主要方法:
- 开发了PH-LLM (Public Health Large Language Models for Infoveillance),这是一套在来自36个数据集的593,100个指令-输出对的多语言集体上训练的LLM.
- 使用量化低级适配器 (QLoRA) 和LoRA加Qwen 2.5,支持29种语言,跨越6种模型大小 (0.5B到32B).
- 评估PH-LLM使用19个英语和20个多语言公共卫生任务的基准和52,158个未见的社交媒体数据对,与领先的开源和专有模型进行比较.
主要成果:
- 在19个英语和20个多语言任务中,PH-LLM在19个英语和20个多语言任务中始终超过了类似和更大的基线模型.
- PH-LLM-32B取得了最先进的结果,PH-LLM-14B和PH-LLM-32B超越了Qwen2.5-72B-Instruct,Llama-3.1-70B-Instruct,Mistral-Large-Instruct-2407和GPT-4o.等模型的性能.
- PH-LLM-7B表现出竞争力的表现,在英语任务中表现优于几款领先型号,尽管平均得分略低于Qwen2.5-7B-Instruct.
结论:
- PH-LLM提供了实时公共卫生信息监控的重大进步,具有最先进的多语言能力.
- 这些模型为通过社交媒体监测公众对健康问题的情绪提供了具有成本效益的解决方案.
- 在危机期间,PH-LLM可以加强机构的快速响应战略,政策制定和公共卫生沟通.
更多相关视频
相关概念视频
Steps in Outbreak Investigation
102
In the ever-evolving field of public health, statistical analysis serves as a cornerstone for understanding and managing disease outbreaks. By leveraging various statistical tools, health professionals can predict potential outbreaks, analyze ongoing situations, and devise effective responses to mitigate impact. For that to happen, there are a few possible stages of the analysis:
102
Principles of Disease Surveillance
65
Disease surveillance is the systematic collection, analysis, and interpretation of health data essential to the planning, implementation, and evaluation of public health practice. This process integrates data dissemination to entities responsible for preventing and controlling disease, injury, and disability. Surveillance systems provide crucial information for action, helping public health authorities make informed decisions to manage and prevent outbreaks, ensure public safety, optimize...
65
Models of Health Promotion and Illness Prevention I
2.0K
A model is a theoretical way to understand a concept or an idea. Models can overcome barriers to health regardless of diverse economic and cultural backgrounds. In addition, models make the task easier by providing different ways to approach complex issues. There are two major health promotion models: the health belief model and the health promotion model.
The health belief model (HBM) attempts to predict health-related behavior in specific belief patterns. According to the HBM, a person's...
The health belief model (HBM) attempts to predict health-related behavior in specific belief patterns. According to the HBM, a person's...
2.0K
Statistical Methods for Analyzing Epidemiological Data
284
Epidemiological data primarily involves information on specific populations' occurrence, distribution, and determinants of health and diseases. This data is crucial for understanding disease patterns and impacts, aiding public health decision-making and disease prevention strategies. The analysis of epidemiological data employs various statistical methods to interpret health-related data effectively. Here are some commonly used methods:
284
Censoring Survival Data
56
Survival analysis is a statistical method used to analyze time-to-event data, often employed in fields such as medicine, engineering, and social sciences. One of the key challenges in survival analysis is dealing with incomplete data, a phenomenon known as "censoring." Censoring occurs when the event of interest (such as death, relapse, or system failure) has not occurred for some individuals by the end of the study period or is otherwise unobservable, and it might have many different...
56
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
54
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
54


