Scholarlink

<p>The development of neurocognitive biomarkers for schizophrenia (SCZ) has relied on lengthy test batteries that are infeasible to deploy in clinical settings. Using machine learning, we sought to identify a subset of neurocognitive domains that could distinguish between patients with SCZ and healthy comparison subjects (HCS). Leveraging data from 559 patients with SCZ or schizoaffective disorder and 745 HCS who completed 15 neurocognitive assessments spanning a diverse range of neurocognitive domains, we developed a machine learning model that could accurately separate SCZ from HCS (area under the receiver operating characteristic curve of 0.899), and was replicated in an independent cohort. Recursive feature elimination revealed that just two neurocognitive domains—verbal learning and emotion identification—were sufficient to achieve the same classification accuracy. These findings support a ‘less-is-more’ approach to efficient neurocognitive profiling across the schizophreniform spectrum and highlight what may be the most impaired neurocognitive domains in this debilitating disorder.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning enables efficient neurocognitive profiling in patients with schizophrenia

  • 期刊论文

  • 发表日期：

    2026年1月7日

查看全文
  • Robert Y. Chen,
  • Tiffany A. Greenwood,
  • David L. Braff,
  • Laura C. Lazzeroni,
  • Neal R. Swerdlow,
  • Monica E. Calkins,
  • Robert Freedman,
  • Michael F. Green,
  • Ruben C. Gur,
  • Raquel E. Gur,
  • Keith H. Nuechterlein,
  • Allen D. Radant,
  • Jeremy M. Silverman,
  • William S. Stone,
  • Catherine A. Sugar,
  • Ming T. Tsuang,
  • Bruce I. Turetsky,
  • Gregory A. Light,
  • Debby W. Tsuang

摘要

The development of neurocognitive biomarkers for schizophrenia (SCZ) has relied on lengthy test batteries that are infeasible to deploy in clinical settings. Using machine learning, we sought to identify a subset of neurocognitive domains that could distinguish between patients with SCZ and healthy comparison subjects (HCS). Leveraging data from 559 patients with SCZ or schizoaffective disorder and 745 HCS who completed 15 neurocognitive assessments spanning a diverse range of neurocognitive domains, we developed a machine learning model that could accurately separate SCZ from HCS (area under the receiver operating characteristic curve of 0.899), and was replicated in an independent cohort. Recursive feature elimination revealed that just two neurocognitive domains—verbal learning and emotion identification—were sufficient to achieve the same classification accuracy. These findings support a ‘less-is-more’ approach to efficient neurocognitive profiling across the schizophreniform spectrum and highlight what may be the most impaired neurocognitive domains in this debilitating disorder.