Surfacing Suicidal Risk Through Simulated Social Interaction: Per-Person Language Model Agents as Communicative Stress Tests
shao, w.; Ammerman, B.; Jacobucci, R.
Show abstract
Suicidal risk may be encoded in everyday communication patterns but diluted in routine digital interactions. We introduce a method for surfacing this latent signal: training per-person language model agents on individuals' authored text (the on-screen text each participant typed, captured whenever a keyboard was visible in screenshots) and placing those agents in simulated social interactionsa communicative stress test. Using data from 79 adults with recent suicidal ideation, we ne-tuned individual LoRA adapters on Qwen3-8B using each participant's authored text, then placed agents in standardized conversations with probe personas. Agent-generated risk language was associated with EMA-measured suicidal ideation (r= .576, p < .001), with a single neutral small-talk probe performing nearly as well (r= 551). A shue control conrmed the signal is person-specic (r= .071 when adapters were mismatched), and automated descriptions of participants' general smartphone activity produced no signal, conrming specicity to interpersonal communication. A prompt ablation demonstrated partial robustness to removal of disclosure-encouraging language (r = .430). This proof-of-concept demonstrates that simulated social interaction can amplify latent vulnerability signals, bridging digital phenotyping, generative AI, andsuicide theory.
Matching journals
The top 10 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- A Comparison of Pruning During Multi-Step Planning in Depressed and Healthy Individuals 90%
- Robotically-induced auditory-verbal hallucinations: combining self-monitoring and strong perceptual priors 90%
- Predicting involuntary admission following inpatient psychiatric treatment using machine learning trained on electronic health record data 90%
Similar papers in this journal
Similar papers in this journal
- Computational Mechanisms of Approach-Avoidance Conflict Predictively Differentiate Between Affective and Substance Use Disorders 92%
- Risk and loss aversion and attitude to COVID and vaccines in anxious individuals 90%
- Computational modelling reveals slower safety learning and threat extinction are associated with higher anxiety severity in remote fear conditioning 90%
Similar papers in this journal
- Individuals with Methamphetamine Use Disorder Show Reduced Directed Exploration and Learning Rates Independent of an Aversive Interoceptive State Induction 91%
- The Temporal Dynamics of Metacognitive Experiences Track Rational Adaptations in Task Performance 89%
- Cognitive Computational Model Reveals Repetition Bias in a Sequential Decision-Making Task 88%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.