Skip to main content
KBS_Icon_questionmark link-ico

Using AI to explore the Hikikomori Phenomenon: An international infodemiology study using social media

Dr. Mariana Pinto da Costa, a Senior Lecturer at the Institute of Psychiatry, Psychology & Neuroscience in the Department of Psychological Medicine, discusses hikikomori, a phenomenon marked by severe social isolation. She provides an overview of an infodemiology study she co-authored, which entailed analysing posts on the social media platform X (formerly known as Twitter) using Artificial Intelligence.

In recent years, researchers have turned to X/Twitter as a valuable tool for investigating various phenomena. It offers a unique space to reach individuals with particular characteristics or experiences, seeking their perspectives, that might otherwise be marginalised or excluded. Social media platforms like this offer a dynamic landscape which is particularly useful for studying the hikikomori phenomenon, given that those affected often seek solace in online social media.

Infodemiology, a burgeoning scientific field, delves into collecting, analysing and interpreting health-related information sourced from electronic media like the internet and other digital sources. We used this approach as it can offer valuable insights that can inform public health strategies and policies.

In today’s digital age, where the internet serves as a hub for sharing opinions, knowledge and problems, platforms like this provide us with novel avenues for communication and information dissemination.

What is Hikikomori?

Hikikomori, originating as a Japanese concept, describes an extreme form of social withdrawal where individuals isolate themselves at home for extended periods, often lasting six months or more. This withdrawal involves a disengagement from education, work, and other daily activities, resulting in significant distress or functional impairment. While initially associated with Japanese culture, studies have revealed its presence in diverse cultural contexts worldwide.

Understanding hikikomori on social media 

Our study investigated the discourse surrounding hikikomori on X/Twitter within Portuguese-speaking users. With Portuguese being one of the world’s most spoken languages, spanning continents like Europe, America, Africa and Asia, our research sought to provide a comprehensive understanding of this phenomenon within Portuguese-speaking populations.

We analysed 13,915 publicly available tweets posted in the Portuguese language containing the keyword "hikikomori" over nearly 15 years. We characterised each tweet using a predefined codebook, considering aspects such as tone, content, user types, and geographical distribution. We examined whether content was from a positive (tweets that express solidarity, self-disclosure, encouragement, gratitude, enthusiasm and pride) or negative perspective (tweets that express blame, stigma, and negative opinions).

We first manually classified a sub-set of posts. We then used artificial intelligence to evaluate tweets. We used a pre-trained neural network called BERTWEET trained on 850 million English tweets to classify hikikomori-related tweets into these different categories.

Key findings

  • Interest and Engagement: There was a high number of likes (N=21,307) and retweets (N=5,359) of hikikomori-related tweets, with a higher number after the pandemic. This underscores the growing awareness and concern regarding social withdrawal and isolation.
  • Content and Tone: Tweets predominantly portrayed hikikomori in a negative tone (N=10,731, 77.1%), with many focusing on curiosities (N=9,818, 70.6%) about the phenomenon. This highlights the prevalent perception of hikikomori as a concerning and undesirable state.
  • User Types: Most users were individuals with hikikomori (N=5,571, 40%). Different user types approached the topic differently, with individuals with hikikomori sharing personal stories (N=3,536, 96.2%), while family and friends posted mainly tweets on curiosities (N=4,526, 46.1%) seeking to understand the phenomenon.
  • Emotional Analysis: While many tweets exhibited a neutral tone (N=4,991, 35.9%), the presence of disgust (N=273, 2.0%) as the least frequent emotion underscores the complexity and discomfort surrounding hikikomori.
  • Geographical Distribution: Tweets originated from different continents, indicating a global awareness of hikikomori among Portuguese speakers.

And now what?

Our study offers invaluable insights into the attitudes, content, and engagement surrounding hikikomori among users of this social media platform.  The hikikomori phenomenon is actively discussed among Portuguese speakers spanning continents, demonstrating its widespread prevalence and the nuances of its globalisation. These findings hold promise in fostering support networks, spearheading awareness campaigns, targeted interventions and support mechanisms to assist those grappling with the challenges of hikikomori. 

The expansive reach of social media offers a unique opportunity to identify and connect with individuals who may be socially withdrawn, providing a platform to not only identify potential risks but also to provide support. Understanding the pivotal role of platforms like X/Twitter in promoting mental health is paramount to developing effective support systems. By harnessing the power of social media platforms, we can disseminate crucial information and facilitate access to resources for those in need, thereby bolstering mental health and wellbeing within our digital communities.

Looking ahead, further research endeavours could investigate this phenomenon in other social media platforms, offering a more comprehensive understanding of the global impact of hikikomori across diverse cultural contexts.

Analysis of the hikikomori phenomenon – an international infodemiology study of Twitter data in Portuguese, ( (Correia Lopes, F., Pinto da Costa, M., Fernandez-Lazaro, C.I. et al.) was published in BMC Public Health. 


In this story

Mariana  Pinto da Costa

Mariana Pinto da Costa

Senior Lecturer

Latest news