{
  "id": 6912847,
  "title": "How AI can help with early schizophrenia diagnosis",
  "url": "https://urgent.news/2026/09/12/how-ai-can-help-with-early-schizophrenia-diagnosis",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-12T11:00:00.000Z",
  "source": {
    "name": "Scientific American",
    "slug": "scientific-american",
    "url": "https://www.scientificamerican.com/article/how-ai-can-help-with-early-schizophrenia-diagnosis/"
  },
  "original_language": "en",
  "account": "Schizophrenia can be challenging to diagnose, with patients exhibiting symptoms like hallucinations, social withdrawal, and delusions in varying degrees. Clinicians often rely on subjective assessments to determine the severity of symptoms and make a diagnosis, leading to delays and inconsistencies in diagnosis. In the United States, individuals with psychotic disorders, including schizophrenia, are typically diagnosed approximately 18 months after their initial symptoms emerge.\n\nResearchers are exploring whether artificial intelligence (AI) could improve schizophrenia diagnosis and care by analyzing subtle differences in speech that might be overlooked by human psychiatrists. While AI is not yet widely implemented in clinics, researchers believe it has the potential to enable early, accurate detection and personalized monitoring of mental illnesses based on subtle symptoms.\n\nA team of researchers in the Netherlands trained an AI system on audio recordings of individuals previously diagnosed with schizophrenia, analyzing 88 different speech features such as loudness, pausing, vowel pronunciation, and intonation. The AI achieved an 86.2 percent accuracy rate in distinguishing between those with and without schizophrenia, even in distinguishing between different subtypes of the disorder. Although this method may not be particularly useful for clear cases, it has the potential to detect early-stage symptoms or high-risk patients and predict relapse.\n\nAnother study focused on the content of speech rather than its sounds. Researchers built a machine learning model to assign a unique \"address\" to each word in a patient's conversation transcript. By examining the patterns of these addresses, the AI could determine whether a sentence remained within the same general topic or shifted to a new subject, indicating disorganized thinking. In this study, the AI achieved an 87 percent accuracy rate in distinguishing between people with and without schizophrenia, outperforming clinical raters who were only 68 percent accurate. Ultimately, AI may play a role in routine symptom monitoring over time, helping clinicians provide more precise and tailored care for patients with schizophrenia.",
  "summary": "The speech of people with schizophrenia sounds subtly different from their healthy counterparts. Can AI hear those differences better than human psychiatrists?",
  "key_points": [],
  "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."
}