Pediatric Sepsis-AKI Urinary Metabolomics Data from a Two-Center Study

360 pediatric urine samples analyzed via GC-MS form the basis for a machine learning model predicting sepsis-associated acute kidney injury. The support vector machine model achieved an AUC of 0.94 in the discovery cohort and 0.89 in external validation. Yali Qian published this prospective observational study data under a CC-BY-4.0 license in April 2026.

Use Cases

• Training machine learning classifiers for early sepsis-AKI prediction based on urinary metabolic fingerprints.
• Validating biomarker panels for pediatric acute kidney injury using metabolomics data.
• Developing SHAP-based model explanations for clinical decision support tools.
• Benchmarking SVM and other ML algorithms on small-sample, high-dimensional clinical metabolomics data.

Strengths

• Data originates from a two-center prospective observational study involving 360 pediatric patients.
• The derived SVM model demonstrated high predictive performance with AUCs of 0.94 and 0.89 in discovery and validation cohorts.
• A panel of 10 discriminative metabolites was identified and integrated into an open-access online platform.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 26.9 KB, indicating limited raw data scope.

Provenance

Source

figshare, author Yali Qian.

Collection Method

Prospective observational study; urine samples analyzed using gas chromatography-mass spectrometry (GC-MS).

Freshness

Last updated 2026-04-22 06:13:37; freshness should be verified.

Geography

Data collected from two clinical centers; specific locations unknown.