Alcohol, media, and emerging technology research (CAPR)

🔔 This week in InSight+

Why we must invest now in digital health interventions for adolescents

👉 buff.ly/EI5UgRA

Reducing alcohol advertising is one of the WHOs best buy policy recommendations. Young people shouldn't see or be targeted with alcohol advertising online and more should be done to reduce exposure.

Most alarmingly, we set up a simulated account for a user under 15 and found that 98% of the posts were visible to those who cannot legally purchase alcohol.

We found that over a third referenced their brand in their bio, that over 75% referenced their brand in at least one post, and that fewer than 2% were flagged as advertising.

We identified 112 celebrities who owned an alcohol brand and had an active Instagram account. We analysed the content in their Instagram bios and collected all of their public posts (over 85k) to determine whether they mentioned their brand.

New paper published in Pediatrics @ameracadpeds.bsky.social, led by PhD student Gedefaw Alen: "Celebrity-Owned Alcohol Brands Promoted on Social Media" doi.org/10.1542/peds...

Such a cool paper! @kbackholer.bsky.social report results from their work on analysing social media data to detect alcohol marketing.

🆕 Cannabis Marketing Restrictions and Exposure to Cannabis Marketing in Legal US Cannabis Markets: Findings From the International Cannabis Policy Study

Lauren Winfield-Ward, Elle Wadsworth, Pete Driezen, David Hammond

doi.org/10.1111/dar....

@uwaterloo.ca @randeurope.org @uniofbath.bsky.social

When alcohol fills the gap: Basic Psychological Need Compensation Predicts Hazardous Drinking

@jconigr.bsky.social on chronic frustration through the lens of Self-Determination Theory

For researchers aiming to analyse large video or image datasets, we also have a number of tutorials on how these models can be used: psycnet.apa.org/fulltext/202...

onlinelibrary.wiley.com/doi/full/10....

These findings go very nicely with Samatha's recent paper in @apsad-dar.bsky.social, which focused on the optimal number of frames to analyse in movies to most accurately estimate how common alcohol is in popular films. onlinelibrary.wiley.com/doi/10.1111/...

The findings can help researchers and policy makers who are aiming to estimate how common alcohol is in video data. This may be particularly important for helping with movie classifications or determining whether social media posts are appropriate for younger viewers.

a close up of lava flowing through a rocky surface ALT: a close up of lava flowing through a rocky surface

The paper found that AI models appear to be able to detect alcohol imagery in videos at large scale with high accuracy and in near real time. Of the three AI models tested, using zero-shot learning with LLaVA achieved the best balance between accuracy and speed.

New paper from @samathasalim.bsky.social published in @addictionjournal.bsky.social: Comparing the accuracy of artificial intelligence models to detect alcohol in video images. onlinelibrary.wiley.com/doi/10.1111/...

2) find alternative activities (e.g., an activity or hobby with no drinking)

3) find social circles or events where others are not drinking.

Feeling like you are missing out when stopping/reducing drinking can be a tough, especially when alcohol use is social. The most common tips from Redditors were:
1) Focus on the positives or what you gain (Joy of missing out)

New paper on FoMO! We analysed Reddit posts to determine how people who were aiming to reduce their alcohol use dealt with the feelings of FoMO. psycnet.apa.org/fulltext/202...

In this analysis, @samathasalim.bsky.social used zero-shot learning and our team have published tutorials aimed at social scientists so that these models are more accessible. psycnet.apa.org/fulltext/202...

The key finding is that we don't need to analyse every frame of a film (25 frames per second). We can instead analyse just 1 frame per second, which is 25x faster and has very little impact on our estimates of how much alcohol is in a film.

A Scalable Sampling Approach for Artificial Intelligence‐Based Alcohol Content Estimation in Movies Introduction The growing accessibility of movies through streaming platforms has expanded audience reach, but also increases exposure to alcohol portrayals, which is an established risk factor for a.....

New paper from our PhD student @samathasalim.bsky.social! The paper aimed to determine how we can more efficiently use AI to analyse movies to estimate how often alcohol is shown. onlinelibrary.wiley.com/doi/10.1111/...

New online: A Scalable Sampling Approach for Artificial Intelligence-Based #Alcohol Content Estimation in #Movies

Samatha Pararath Salim, Zhen He, Emmanuel Kuntsche, Joshua Millward, Benjamin Riordan

doi.org/10.1111/dar....

@latrobeuni.bsky.social @alcoholandmedia.bsky.social #AI

@apadivision50.bsky.social @apadiv28.bsky.social @apsad.bsky.social @drinkingstudies.bsky.social @latrobeuni.bsky.social @uconnmhealth.bsky.social @capraustralia.bsky.social

The tutorial includes 135 images of the lead author posing with different beverages and we walk you through how to use zero-shot learning to analyse the image to determine what beverage is seen and whether the beverage is alcoholic or not.

Google Colab

All of the code can be run on GoogleColab, so you don't need to download any programs and can just click your way through and start analysing some images. colab.research.google.com/github/ltu-c...

New open access tutorial paper on how social scientists can use machine learning to analyse images. We think this is a super cool method and we have aimed it at those with no background in programming or computer science psycnet.apa.org/fulltext/202...

We need better measures to understand the influence of social media on substance use Click on the article title to read more.

We need better measures if we want to properly understand the impact of social media use or exposure to alcohol-related content on alcohol use.
onlinelibrary.wiley.com/doi/full/10....

Sage Journals: Discover world-class research Subscription and open access journals from Sage, the world's leading independent academic publisher.

“I Don’t Know for Certain”: A Content Analysis of Reddit Posters’ Accounts of Drink Spiking

New paper looking at relationships, contexts and impacts of alcohol & drug-facilitated sexual violence ('drink spiking'). Led by Jess Ison & @alcoholandmedia.bsky.social

journals.sagepub.com/doi/10.1177/...

How to apply zero‐shot learning to text data in substance use research: An overview and tutorial with media data A vast amount of media-related text data is generated daily in the form of social media posts, news stories or academic articles. These text data provide opportunities for researchers to analyse and ...

Professor Kuntsche drew on some of his recent papers including his chapter in the Handbook of Addiction Psychology and some of our recent tutorial papers. onlinelibrary.wiley.com/doi/full/10....

us.sagepub.com/en-us/nam/th...

CAPR's director Emmanuel Kuntsche rocked the mic at the recent Global Congress on Behavioural Addictions in Istanbul. Here are some pics from his presentation on how machine learning is used in addiction research.

Examining gender in alcohol research: A systematic review of gender differences in how men and women are studied in alcohol research This systematic review examined differences in the way women and men have been studied in alcohol research over the past decade. In particular, it exp…

Their massive systematic review was recently published in The International Journal of Drug Policy Research: www.sciencedirect.com/science/arti...