Scholarlink

<p>Schizophrenia (SCZ) is a complex psychiatric disorder, and its pathogenic mechanisms are not yet fully understood. The identification of reliable blood biomarkers and molecular subtypes for early diagnosis and effective therapy remains a significant challenge. To address this issue, we utilized a combination of bioinformatics and machine learning (ML) to identify potential biomarkers for SCZ. Our approach involved the integration of 12 different ML algorithms to develop a diagnostic signature based on data from several datasets, including GSE18312, GSE27383, GSE38485, GSE54913, and GSE165604. A nomogram was constructed using these datasets for potential clinical applications. In addition, clustering analysis was performed on SCZ patients using consensus clustering and non-negative matrix factorization (NMF) algorithms. We further evaluated subtype differences in biological functions and immune cells through various methods, such as gene set enrichment analysis (GSEA), gene set variation analysis (GSVA), Proteomaps, and IOBR analyses. Our results identified a diagnostic signature composed of 16 genes (APBB2, CLCN1, SYDE1, PAX5, SNAI1, DAZL, UNC93B1, PLAGL2, HS3ST1, ITPKB, PILRA, BTLA, SWAP70, AZI2, ADM, and AVPR2), which demonstrated robust performance in diagnosing SCZ across eight different datasets. A nomogram based on these genes was created, providing clinical benefits for SCZ patients. Among the identified genes, AZI2 was found to be the most critical, influencing inflammation and immunity. We also identified potential chemical compounds that could target these 16 genes. Unsupervised clustering and NMF algorithms revealed two distinct subtypes of SCZ, each associated with unique immune cell profiles, biological functions, and protein expression levels. In conclusion, this study not only developed a diagnostic signature and a novel nomogram for SCZ but also provided new insights into the subtypes of SCZ. These findings may pave the way for personalized diagnosis and treatment strategies for SCZ patients.</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-based predictive models and subtypes patterns in peripheral blood of schizophrenia based on a machine learning computational framework

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月24日

查看全文
  • Zhijun Li,
  • Qing Sun,
  • Haoyu Li,
  • Naiyu Guan,
  • Jing Ni,
  • Jing Wang,
  • Xiaolei Xu,
  • Ye Shen,
  • Siyu Sun,
  • Yan Li

摘要

Schizophrenia (SCZ) is a complex psychiatric disorder, and its pathogenic mechanisms are not yet fully understood. The identification of reliable blood biomarkers and molecular subtypes for early diagnosis and effective therapy remains a significant challenge. To address this issue, we utilized a combination of bioinformatics and machine learning (ML) to identify potential biomarkers for SCZ. Our approach involved the integration of 12 different ML algorithms to develop a diagnostic signature based on data from several datasets, including GSE18312, GSE27383, GSE38485, GSE54913, and GSE165604. A nomogram was constructed using these datasets for potential clinical applications. In addition, clustering analysis was performed on SCZ patients using consensus clustering and non-negative matrix factorization (NMF) algorithms. We further evaluated subtype differences in biological functions and immune cells through various methods, such as gene set enrichment analysis (GSEA), gene set variation analysis (GSVA), Proteomaps, and IOBR analyses. Our results identified a diagnostic signature composed of 16 genes (APBB2, CLCN1, SYDE1, PAX5, SNAI1, DAZL, UNC93B1, PLAGL2, HS3ST1, ITPKB, PILRA, BTLA, SWAP70, AZI2, ADM, and AVPR2), which demonstrated robust performance in diagnosing SCZ across eight different datasets. A nomogram based on these genes was created, providing clinical benefits for SCZ patients. Among the identified genes, AZI2 was found to be the most critical, influencing inflammation and immunity. We also identified potential chemical compounds that could target these 16 genes. Unsupervised clustering and NMF algorithms revealed two distinct subtypes of SCZ, each associated with unique immune cell profiles, biological functions, and protein expression levels. In conclusion, this study not only developed a diagnostic signature and a novel nomogram for SCZ but also provided new insights into the subtypes of SCZ. These findings may pave the way for personalized diagnosis and treatment strategies for SCZ patients.