Data Sheet 1_Inferring tumor immune microenvironment -related risk states from pretreatmen

Lei Yuan's dataset from a multicenter study includes pretreatment H&E whole slide images and clinical variables from 346 eligible patients with advanced non-small cell lung cancer (NSCLC) screened between January 2022 and January 2025. The data was used to develop and test multimodal AI models for predicting the risk of checkpoint inhibitor pneumonitis (CIP). The study demonstrates that a fusion model integrating pathomics and clinical biomarkers achieved high predictive performance, with AUC scores up to 0.930.

Use Cases

• Training multimodal AI models to predict checkpoint inhibitor pneumonitis risk based on H&E whole slide images and clinical biomarkers.
• Benchmarking pathomics feature extraction methods against clinical variable models for adverse event prediction.
• Developing cross-modal representation learning frameworks for integrating image and tabular clinical data in oncology.
• Analyzing the tumor immune microenvironment from routine H&E pathology slides using convolutional neural networks and multi-instance learning.

Strengths

• Multicenter study design involving 346 eligible patients from three hospitals.
• Multimodal data integration combining H&E whole slide images with structured clinical variables.
• Model performance metrics are provided, with the fusion model achieving AUC scores of 0.930, 0.919, and 0.905 across validation and test sets.
• Data is shared under a permissive CC-BY-4.0 license.

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 small in file size (493.4 KB), suggesting it may contain metadata or summary results rather than the raw image data.

Provenance

Source

figshare, author Lei Yuan.

Collection Method

Retrospective study screening patients consecutively from three hospitals.

Time Range

January 2022 to January 2025.

Freshness

Last updated 2026-03-17 13:06:59.

Geography

Multicenter study; specific geographic locations are not stated.