Scholarlink

Background <p>Sepsis remains a life-threatening condition in intensive care units(ICU) with high morbidity and mortality rates. Some biomarkers commonly used inclinic do not have the characteristics of rapid and specific growth and rapiddecline after effective treatment. Machine learning has shown great potential inearly diagnosis, subtype analysis, accurate treatment and prognosis evaluationof sepsis.</p> Methods <p>Gene expression matrices from GSE13904 and GSE26440 were combinedinto a training model after quality control and standardization. Then, theintersection genes were obtained by crossing the screened differentiallyexpressed genes (DEGs) and the module genes with the strongest correlationobtained by WGCNA analysis. 113 combined machine learning algorithms to build adiagnosis model. Then the CIBERSORT algorithm is used to analyze therelationship between the change of core gene expression and immune response insepsis. Construct nomogram, DCA and CIC to further verify the reliability of thediagnosis model. The potential molecular compounds interacting with key geneswere searched from the Traditional Chinese Medicine Active Compound Library(TCMACL).</p> Results <p>We screened 405 DEGs, including 334 up-regulated and 71down-regulated genes. The 308 potential genes were obtained by intersection ofMEturquoise module genes in WGCNA analysis and DEGs for subsequent machinelearning analysis. GO and KEGG enrichment analysis showed that sepsis was mainlyrelated to immune response and bacterial infection. Then 113 combined machinelearning algorithms are applied to construct a diagnosis model to screen 22 hubgenes. Four four key genes (CD177, GNLY, ANKRD22, and IFIT1) are obtainedthrough further analysis of PPI network constructed by 22 hub genes.Subsequently, the diagnostic model is proved to have good predictive value bynomogram, DCA and CIC. Finally, molecular compounds (Dieckol, Grosvenorine andTellimagrandin II) were screened out as potential drugs.</p> Conclusion <p>113 combinated machine learning algorithms screened out four keygenes that can distinguish sepsis patients. At the same time, potentialtherapeutic molecular compounds interacting with key genes genes were screenedout by molecular docking.</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

RETRACTED ARTICLE: A diagnostic model for sepsis using an integrated machine learning framework approach and its therapeutic drug discovery

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月14日

查看全文
  • Wuping Zhang,
  • Hanping Shi,
  • Jie Peng

摘要

Background

Sepsis remains a life-threatening condition in intensive care units(ICU) with high morbidity and mortality rates. Some biomarkers commonly used inclinic do not have the characteristics of rapid and specific growth and rapiddecline after effective treatment. Machine learning has shown great potential inearly diagnosis, subtype analysis, accurate treatment and prognosis evaluationof sepsis.

Methods

Gene expression matrices from GSE13904 and GSE26440 were combinedinto a training model after quality control and standardization. Then, theintersection genes were obtained by crossing the screened differentiallyexpressed genes (DEGs) and the module genes with the strongest correlationobtained by WGCNA analysis. 113 combined machine learning algorithms to build adiagnosis model. Then the CIBERSORT algorithm is used to analyze therelationship between the change of core gene expression and immune response insepsis. Construct nomogram, DCA and CIC to further verify the reliability of thediagnosis model. The potential molecular compounds interacting with key geneswere searched from the Traditional Chinese Medicine Active Compound Library(TCMACL).

Results

We screened 405 DEGs, including 334 up-regulated and 71down-regulated genes. The 308 potential genes were obtained by intersection ofMEturquoise module genes in WGCNA analysis and DEGs for subsequent machinelearning analysis. GO and KEGG enrichment analysis showed that sepsis was mainlyrelated to immune response and bacterial infection. Then 113 combined machinelearning algorithms are applied to construct a diagnosis model to screen 22 hubgenes. Four four key genes (CD177, GNLY, ANKRD22, and IFIT1) are obtainedthrough further analysis of PPI network constructed by 22 hub genes.Subsequently, the diagnostic model is proved to have good predictive value bynomogram, DCA and CIC. Finally, molecular compounds (Dieckol, Grosvenorine andTellimagrandin II) were screened out as potential drugs.

Conclusion

113 combinated machine learning algorithms screened out four keygenes that can distinguish sepsis patients. At the same time, potentialtherapeutic molecular compounds interacting with key genes genes were screenedout by molecular docking.