People First Radio – July 18, 2024People First Radio – July 18, 2024
People First Radio
The episode looks at research into using social media language patterns as a potential aid for detecting psychosis, alongside first-hand accounts of psychosis and schizophrenia. It raises ethical concerns about privacy and stigma while highlighting recovery, support, and the need for more open public discussion.
0:00•18 Jul 2024
Can Your Social Media Posts Hint at Psychosis?
Episode Overview
- Language models can measure speech coherence in Reddit posts, and lower coherence appears more often in psychosis-related forums than general forums.
- Laboratory studies of speech have reached high accuracy in identifying schizophrenia and even predicting future onset, but social media-based systems are currently less accurate.
- Using social media for early identification could widen access and reduce costs, yet raises serious risks around privacy, discrimination, and employer misuse.
- Lived experiences of psychosis and schizophrenia show how frightening symptoms, isolation, and lack of understanding can be, while also highlighting the value of medication, CBT techniques, and patient support.
- Stigma affects housing, employment, and language, and guests argue for person-first terminology and stronger allyship for people living with schizophrenia.
“The most important point is what you say online really stays online and can be used for various things.”
What insights can experts and survivors share about addiction and mental health? This instalment of People First Radio takes on that question by looking at how social media might hint at serious conditions like psychosis, while centring the voices of people who have lived through it.
Host Joe Pugh talks with German PhD student Loren Planck about research into “detecting psychosis from social media posts.” Planck explains how language models similar to chatbots are used to measure “speech coherence” in Reddit posts from psychosis-related forums compared with other communities.
As he puts it, “we could maybe someday use social media data to predict a mental disorder or to indicate that a person might be suffering from a mental disorder.” You’ll hear about the potential promise of earlier detection for conditions such as schizophrenia, especially in a world where millions struggle to access timely mental health care.
At the same time, the conversation raises tough questions: who owns this data, who gets to use it, and what happens if employers or other organisations quietly run these checks in the background? Planck stresses that “what you say online really stays online and can be used… for various things,” highlighting the need for public debate and clear safeguards. To keep the discussion grounded, Joe also shares excerpts from earlier interviews.
Student Gurleen Kaur describes being “literally locked in my room for days” during psychosis and how medication, CBT tools, and support helped her slowly rebuild life.
Writer and teacher Leif Gregerson talks about living with schizophrenia, the impact of hospitalisation, and how stigma shows up in housing, work, and everyday language—arguing for people-first terms like “person living with schizophrenia.” If you’re curious about how technology, ethics, and real human stories collide in modern mental health care, this is a thoughtful listen that might leave you asking: how should we use the digital traces we all leave behind?
Do you want to link to this podcast?
Get the buttons here!
More From This Show
The latest episodes from the same podcast.