Digital Monitoring Tracks Suicidal Thoughts After Hospital Discharge

The period immediately after psychiatric hospital discharge represents one of the highest-risk times for suicide, with rates spiking up to 200 times higher than baseline. Traditional monitoring through occasional outpatient visits fails to capture the rapid fluctuations in suicidal thoughts that can occur throughout a single day.

The MULTICAST-PREDICT study introduces a revolutionary "deep phenotyping" approach to address this critical gap. Researchers will recruit 200 patients with current or past suicidal thoughts and behaviors across psychiatric hospitals in Zurich, Basel, and New York. The comprehensive monitoring system combines multiple data streams: baseline brain scans using EEG to identify neurobiological markers, video-recorded language tasks to analyze speech patterns and syntax, and a smartphone app that tracks patients for 28 days post-discharge.

The smartphone monitoring includes two intensive weeks of five daily mood assessments and video diaries every other day, plus continuous passive sensing of movement, sleep, and phone usage patterns. This multi-modal approach aims to identify early warning signs that precede dangerous increases in suicidal ideation.

Key innovations include using machine learning to integrate psychological, neurobiological, linguistic, and digital behavioral data into predictive models. The researchers hypothesize that reduced syntactic complexity in speech, altered brain wave patterns, and changes in smartphone usage patterns will collectively predict suicide risk more accurately than traditional clinical assessments alone.

If successful, this approach could transform suicide prevention by enabling real-time intervention when patients are most vulnerable, potentially saving thousands of lives annually through early detection and timely clinical response.