YouTube Videos for Public Health Literacy - A Machine Learning Pipeline to Curate Covid-19 Videos

Aug 5, 2026

The COVID-19 pandemic underscored the need to strengthen public health literacy for social resilience. YouTube, as a large, multimedia repository of user-generated health content, can support understanding and learning, but quality and accuracy concerns limit its usefulness. This study presents an automated pipeline to identify, retrieve, and shortlist medically relevant and understandable videos for expert review and public dissemination. The approach combines domain knowledge with machine learning and NLP for scalable, replicable, and generalizable curation of COVID-19 health information and similar outbreaks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

AccessFree to download

Introduction

Public health literacy and the role of social media

Problem: content quality and misinformation

Methods

Data collection and annotation

Keyword generation and video retrieval

Video filtering and licensing constraints

Evaluation

Medical relevance and understandability criteria

YouTube Videos for Public Health Literacy? A Machine Learning Pipeline to Curate Covid-19 Videos Yawen GUOa, Xiao LIUb, Anjana SUSARLA c, Rema PADMANda University of California, Irvine bArizona State University cMichigan State University d Carnegie Mellon University Abstract. The COVID-19 pandemic has highlighted the dire necessity to improve public health literacy for societal resilience. YouTube, the largest video-sharing social media platform, provides a vast repository of user-generated health information in a multi-media-rich format which may be easier for the public to understand and use if major concerns about content quality and accuracy are addressed. This study develops an automated solution to identify, retrieve and shortlist medically relevant and understandable YouTube videos that domain experts can subsequently review and recommend for disseminating and educating the public on the COVID-19 pandemic and similar public health outbreaks. Our approach leverages domain knowledge from human experts and machine learning and natural language processing methods to provide a scalable, replicable, and generalizable approach that can also be applied to enhance the management of many health conditions. Keywords. Visual social media, machine learning, natural language processing, healthcare informatics, COVID-19 literacy 1. Introduction The easy availability of vast amounts of medical information on the Internet, coupled with the rapid growth in the use of social media by the public, patients, and clinicians alike, has transformed how consumers access medical information to manage their health information needs and illnesses [1] . Traditionally, patients receive such information and instructions in text format from their healthcare providers and organizations. However, purely text-based medical information has been found to reduce user attention, understanding, recall, and compliance, especially for patients with low literacy levels [2] . Hence, it is important to design educational materials for patients and the general public that increase engagement and participation in health-related decision-making. YouTube hosts millions of health-related videos about the pathogenesis, diagnosis, treatment, and prevention of many medical conditions, including COVID-19. This vast repository of audio-visual content, created by both professional healthcare organizations and random individuals, is of widely varying quality and a challenge for both the public and healthcare professionals to search and retrieve credible and relevant videos as a justin-time, contextualized, prescriptive, digital therapeutic and educational intervention. This study aims to address the challenge of automating the identification, retrieval and curation of useful health-related YouTube video content via a machine learning pipeline from the perspective of improving public health literacy, using COVID-19 as an illustrative example. Evaluating encoded medical content in a video and its understandability are two critical criteria we apply to assess a recommended video [1] . Studies assessing the readability, relevance, or comprehensibility of patient education materials on a myriad of topics abound. The evidence is clear and consistent that most education materials are too complex for most consumers because many adults lack the requisite skills to obtain and process basic health information and services needed to make appropriate health decisions [3] . Responding to the health communication challenges raised by the pandemic, public health agencies must use resources such as YouTube to better deliver timely and accurate information and to minimize the spread of misinformation. 2. Methods 2.1. Data Collection & Annotation To assess the amount of medical information encoded in YouTube videos and their understandability, we develop a data collection process that first generates keywords about COVID-19 and then retrieves the videos. We select t

FAQs

What problem does the pipeline address for public health literacy?

It targets the difficulty of automatically identifying and curating useful, medically relevant YouTube videos while ensuring the content is understandable and accurate enough for public education.

How are COVID-19 keywords generated for video retrieval?

Keywords are selected from three sources: terms extracted from DailyStrength expert forum posts, YouTube search suggestions under the COVID-19 search bar, and FAQ lists from WHO and CDC. The union of these sources is ranked by popularity to choose the top terms.

What filtering rules are applied to the collected YouTube videos?

Videos are filtered to English language and a viewing-friendly duration range of 60 seconds to 6 minutes. Items that cannot be downloaded due to video license issues are excluded.

Aug 5, 2026

Upload by Genevieve