• Home
  • News
  • Beware of the bots! new study highlights role of bots in the #healthydiet discourse on twitter

Oct 14 2021


Opinion

Beware of the Bots! New study highlights role of bots in the #healthydiet discourse on Twitter

This article reflects the authors own views and not that of MLI.

By Aoife O’Reilly, Communications Manager – Digital and Health safefood

Bots play a significant role in the healthy diet conversation on Twitter, a recent study conducted by The Irish Institute of Digital Business at Dublin City University, in conjunction with safefood, has found.

81% of the 100 most active users were bots-like accounts according to the study “Sorting the Healthy Diet Signal from the Social Media Expert Noise: Preliminary Evidence from the Healthy Diet Discourse on Twitter”.

It recommends that health professionals should actively engage with social media and coordinate with each other to amplify public health messages and cut through the noise created by unqualified accounts.

To establish the most influential users and communities discussing healthy diets, the study analysed over 1.2 million tweets featuring the hashtag #healthydiet generated by nearly 630,000 Twitter accounts from January 2018 to April 2019.

Using a machine learning algorithm to detect social bots on Twitter, the study found that 60 out of the 100 most active users in the #healthydiet discourse had high similarity to bots and other 21 users had since been suspended by the platform. This means that bot-like accounts play a significant role in messaging within the #healthydiet discourse as they generate a disproportionate volume of tweets.

In contrast, the study found that highly visible users, those who received the greatest amount of retweets and replies, had much higher levels of legitimacy. Their Twitter profile and behaviour was considerably less similar to bots. However, they were also significantly less active.

The #healthydiet conversation is dominated by non-health professionals. The study revealed that only 1.1% of the 629,608 accounts analysed had verified accounts. It identified five prominent sub-communities including from largest to smallest:

  1. Special diets such as low-carb, high protein, etc
  2. South Beach diet (@southbeachdiet)
  3. Vegan lifestyle and diets
  4. Traditional online media, publishers and influencers such as Harvard Health Publishing, The Wall Street Journal, Huffington Post and celebrities
  5. Public Health Organisations such as the WHO, The World Economic Forum and public health figures

Diet is the most discussed topic followed by nutrition, exercise, weight, diseases and healthy lifestyle according to the study, which used Health and Ingest dictionaries for its content analysis and a lexicon-based classifier to measure frequency.

High protein/low or no carbohydrate diets (ketogenic diets) forms the most dominant cluster of conversation. A higher proportion of tweets promoting these diets come from bot-like accounts.

Health professionals seeking to influence people’s healthy eating choices face a real challenge when competing for attention in this cluttered and noisy environment. The study recommends that health professionals should:
- actively and consistently engage with social media to help amplify evidence-based information or advice and coordinate with others for impact
- highlight their credentials in their profiles and posts
- share information using conversational and accessible language.

Watch Professor Theo Lynn discussing the findings in a safefood webinar: Exploring the unqualified expert community online.

Read the full report here