HEALTH DISINFORMATION: A SYSTEMATIC LITERATURE REVIEW THROUGH DATA MINING

¹University of Santo Tomas, College of Rehabilitation Sciences, Manila, Philippines, ²Ewha Womans University, Education, Seoul, Korea (Republic of)

Background: It has been predicted that data will be the new oil worldwide. Persons, groups, and organizations accessing and using more data will be advantageous. However, data volume, variety, and velocity increase are highly threatened by the continuously rising prevalence of disinformation in all aspects of life, like health. Understanding the phenomenon of disinformation is relevant to ensure that big data will be used correctly in any field, like health.

Purpose: This study aims to describe the pattern and publication trends in disinformation in health research.

Methods: Research and review abstracts published in English were extracted from Web of Science and Scopus using the keyword (“disinformation” OR “misinformation” OR “fake news” OR “false news” OR “false information” OR “malinformation”) AND (“health”). Data synthesis through pre-processing, word frequency and co-occurrence analysis, topic modeling using Latent Dirichlet Allocation, and trend analysis were done to identify patterns and publication trends. R studio and packages were utilized to manage and analyze the data.

Results: A total of 4972 abstracts were found about the topic. Most publications were related to topics on effects of disinformation (n=696,14.19%), social media (n=657, 13.68), COVID-19 (n=548, 11.41%), women’s health (n=505, 10.52%), and community health (n=459, 9.56%). While topics with the least publications were media and information (n=302, 6.29%), lifestyle and wellbeing (n=353, 7.35%) and vaccination (n=385, 8.02%). Except for the three topics with the least publications (coefficient0.38-0.97,p>0.05), all topics identified in the modeling were positive and significant, indicating increased publications for the past years (coefficient0.52-1.23,p<0.05). Though related publications about health disinformation started as early as the year the 1980s, the majority of all research was published from 2010 onwards.

Conclusions: The use of data mining for health disinformation publications revealed much research on varying topics since the 1980s. Though trends for all topics are generally increasing, variations in publication patterns, which may be related to different factors worldwide, were observed. The phenomenon of disinformation was found to be analyzed and discussed in almost all aspects of health, including vaccination, pandemic, women’s health, media and information, lifestyle and wellbeing, community health, social media, and policies.

Implications: The synthesis and understanding of a large number of research findings in this study may be critical for developing relevant and specific strategies and policies for countering and preventing disinformation in health.

Funding acknowledgements: This study did not receive any grants

Keywords:
Health disinformation
Data mining
Topic modelling

Topics:
Professional practice: other

Did this work require ethics approval? No

Reason: As the study did not involve any human participants, no ethics approval was sought. Source: National Ethical Guidelines for Health and Health-Related Research 2017; Link:https://ethics.healthresearch.ph/index.php/phoca-downloads/category/4-neg)