Multi-Cohort Transcriptomics for a Septic Shock Diagnostic Model

Ling Li's research integrates transcriptomic data from five GEO peripheral blood cohorts to build a diagnostic model for septic shock. The work identifies five signature genes and constructs a 5-gene artificial neural network model, validated across multiple validation cohorts. The dataset, last updated in 2026, includes experimental results elucidating m6A regulatory mechanisms.

Use Cases

• Training a diagnostic classifier for septic shock based on the identified five-gene signature.
• Analyzing neutrophil-specific gene expression patterns using the single-cell data referenced in the description.
• Investigating m6A RNA modification mechanisms in sepsis pathology based on the described METTL14/YTHDF1–S100A12 axis.

Strengths

• Model validation was performed across three independent validation cohorts (GSE95233, GSE131761, and a clinical cohort).
• The diagnostic signature was derived from five distinct GEO cohorts using multiple machine learning algorithms (LASSO, SVM-RFE, random forest).
• The study includes orthogonal validation through single-cell analysis (GSE167363) and in vivo experiments in CLP mice.

Limitations

• The dataset is very small (19.2 KB), suggesting it likely contains summary results or a document rather than raw transcriptomic data.
• Column-level documentation is absent; field semantics must be inferred from the accompanying research document.
• Row count is unknown, which may limit suitability assessment for direct machine learning application.

Provenance

Source

figshare

Collection Method

Integrated analysis of five public GEO transcriptomic cohorts, with subsequent experimental validation.

Freshness

Last updated 2026-05-25 04:34:59