{
  "id": 14228094,
  "title": "Emotion-focused AI spots early signs of psychosis in youths from short speech samples",
  "url": "https://urgent.news/2026/10/13/emotion-focused-ai-spots-early-signs-of-psychosis-in-youths-from",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-13T15:20:01.000Z",
  "source": {
    "name": "Medical Xpress",
    "slug": "medical-xpress",
    "url": "https://medicalxpress.com/news/2026-10-emotion-focused-ai-early-psychosis.html"
  },
  "original_language": "en",
  "account": "Scientists from Nanyang Technological University and the Institute of Mental Health in Singapore have created an AI framework that can detect early signs of psychosis in young people using short, casual speech samples. This research could help primary care providers and community workers identify vulnerable youths before their mental health deteriorates. Psychosis, which often begins during adolescence or early adulthood, involves hallucinations, delusions, and disordered thinking. Early identification is particularly challenging because people rarely openly discuss mental health struggles, a phenomenon known as non-disclosing speech. The study analyzed speech samples from 429 youths aged 14-29 from Singapore's Longitudinal Youth-at-Risk Study. Standard large language models, designed to generate human-like language, underperformed in predicting clinical risk from these transcripts. Researchers developed LEMon-AID, which uses LLMs to detect emotional cues like joy, trust, fear, and sadness, then integrates these emotional signals with traditional machine learning predictions. This approach improved overall accuracy from 55% using conventional methods to 67% with LEMon-AID, and increased detection sensitivity to identify up to 93% of at-risk individuals. LEMon-AID provides clear, understandable explanations for its risk assessments, making it a more transparent and safer tool for clinicians. Dr. Jimmy Lee, who led the study, emphasized that early detection is crucial for preventing long-term disability and restoring the health of young people with mental illness. The researchers plan to further develop LEMon-AID using multimodal approaches, incorporating additional data types and modalities, to enhance its reliability and understanding of mental health progression.",
  "summary": "Nanyang Technological University, Singapore (NTU Singapore) and the Institute of Mental Health (IMH) have developed an artificial intelligence (AI) framework capable of identifying youths at high risk of psychosis from short samples of casual, spontaneous speech.",
  "key_points": [
    "LEMon-AID uses emotional cues from short speech samples to predict psychosis in youths aged 14-29.",
    "Improved accuracy from 55% to 67% and detection sensitivity to 93% of at-risk individuals."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}