De-identified mNGS Report-Derived Pathogen Profiles for Radiomic Subtype Analysis in Pneumonia-Associated Sepsis

This dataset contains de-identified, report-derived pathogen profiles used in the study “Multimodal Deep Learning for Risk Stratification of Imaging Subtypes in Pneumonia-Associated Sepsis.” The data were extracted from clinically ordered hospital metagenomic next-generation sequencing (mNGS) reports. The dataset includes reported pathogen names, relative-abundance estimates, and radiomic subtype.Radiomic subtype labels are included only as derived analytical variables used to reproduce the reported subtype-associated analyses. The dataset does not contain raw FASTQ or BAM files. The study team did not access or reprocess the underlying raw sequencing files, and no independent sequence-level taxonomic calling or quality-control pipeline was performed. The reported sequencing read counts and relative-abundance estimates were obtained from structured hospital laboratory reports. Direct patient identifiers and unnecessary clinical information were removed before deposition. The dataset does not include patient names, medical record numbers, specimen numbers, or exact clinical dates. Users must not attempt to re-identify individuals or link the data with external personal information. The dataset is provided for research and reproducibility purposes. The microbial analyses in the associated study were exploratory and should not be interpreted as evidence of causal microbial interactions.