Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

Cross-reactivity

Cross-reactivity

Vaccinations

Vaccinations

Steps in Outbreak Investigation

Steps in Outbreak Investigation

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:

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

The imaginary divide between mental and "physical" health: Dismantling dualism and reductionism to address a monumental mistake in medicine.

JCPP advances·2026

Same author

Integrating evolutionary and compositional features with ML and DL for robust and interpretable druggable protein prediction.

Journal of computer-aided molecular design·2026

Same author

Enhanced biomethane production associated with olivine-stimulated CO<sub>2</sub> reduction in petroleum reservoir production water.

Bioresource technology·2026

Same author

The Maternal Vaccine Study Protocol: A Victorian Cohort Study Evaluating Infant and Childhood Safety and Health and Developmental Outcomes After Vaccination Against Respiratory Viruses in Pregnancy.

Vaccines·2026

Same author

Delayed diagnosis is associated with complications following invasive meningococcal disease in Australian adolescents and young adults.

Infection·2026

Same author

Maternal preferences for infant respiratory syncytial virus (RSV) preventives: an Australian discrete choice experiment.

Vaccine·2026

Search research articles

Related Experiment Video

Updated: Jun 12, 2025

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Exploring Large Language Models for Detecting Online Vaccine Reactions.

Sedigh Khademi^1,2, Christopher Palmer¹, Gerardo Luis Dimaguila^1,2

¹Murdoch Children's Research Institute, Parkville, Australia.

Studies in Health Technology and Informatics

|September 25, 2024

Summary

Large language models (LLMs) can detect vaccine reactions on social media using prompt engineering. While pretrained language models (PLMs) are more accurate, LLMs aid in identifying data for training PLMs.

Keywords:

large language model natural language processing social media vaccine safety

More Related Videos

Detection of Neutralization-sensitive Epitopes in Antigens Displayed on Virus-Like Particle VLP-Based Vaccines Using a Capture Assay

Detection of Neutralization-sensitive Epitopes in Antigens Displayed on Virus-Like Particle VLP-Based Vaccines Using a Capture Assay

Published on: February 10, 2022

In Vivo Assay for Detection of Antigen-specific T-cell Cytolytic Function Using a Vaccination Model

In Vivo Assay for Detection of Antigen-specific T-cell Cytolytic Function Using a Vaccination Model

Published on: November 28, 2017

Related Experiment Videos

Last Updated: Jun 12, 2025

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Detection of Neutralization-sensitive Epitopes in Antigens Displayed on Virus-Like Particle VLP-Based Vaccines Using a Capture Assay

Detection of Neutralization-sensitive Epitopes in Antigens Displayed on Virus-Like Particle VLP-Based Vaccines Using a Capture Assay

Published on: February 10, 2022

In Vivo Assay for Detection of Antigen-specific T-cell Cytolytic Function Using a Vaccination Model

In Vivo Assay for Detection of Antigen-specific T-cell Cytolytic Function Using a Vaccination Model

Published on: November 28, 2017

Area of Science:

Computational linguistics
Public health informatics
Artificial intelligence in medicine

Background:

Social media generates real-time health data, including adverse vaccine events.
Analyzing this noisy data for specific health insights like vaccine reactions is challenging.

Purpose of the Study:

To evaluate the effectiveness of large language models (LLMs) and prompt engineering for detecting personal mentions of vaccine reactions on social media.
To compare different prompting strategies and LLM models (GPT-3.5, GPT-4) for this task.

Main Methods:

Utilized Reddit data concerning shingles vaccines.
Compared zero-shot and few-shot learning with standard and chain-of-thought prompts across two LLM models.
Assessed LLM performance against lightweight supervised pretrained language models (PLMs).

Main Results:

Chain-of-thought prompts significantly improved LLM identification of relevant social media posts.
Few-shot learning enhanced GPT-4's ability to capture marginal cases, albeit with reduced precision.
Supervised PLMs generally outperformed LLMs in classification accuracy.

Conclusions:

LLMs, particularly with chain-of-thought prompting, show promise for identifying vaccine reaction mentions in social media data.
LLMs can assist in data curation for training PLMs, especially for reducing false negatives.
LLMs serve as viable classifiers when insufficient data prevents training robust PLMs.