<p>Clinical application of the DL model.</p>

1,649 photographs from 615 patients were used to develop and validate a deep learning model for detecting venous congestion in free flap reconstruction between January 2019 and February 2025. The dataset, shared by Xiaoyu Huang on figshare, contains performance metrics from model training, internal validation, and external validation phases. A clinical comparison phase included 113 patients between March and June 2025.

Use Cases

• Benchmarking deep learning model performance for medical image classification based on reported accuracy, sensitivity, and specificity metrics.
• Studying the impact of AI-assisted remote monitoring on clinical outcomes like time to detection and flap survival rates.
• Analyzing the generalizability of a diagnostic model across different validation phases (internal vs. external).

Strengths

• Model performance is documented with specific metrics, including an external validation accuracy of 89.3% and AUC of 0.93.
• Clinical application results are reported, showing an overall accuracy of 92.16% and a false-positive rate of 2.62%.
• Data collection spans a defined multi-year period from January 2019 to June 2025.

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 5.5 KB, suggesting it contains summary statistics rather than raw patient-level data.

Provenance

Source

Xiaoyu Huang

Collection Method

Data was collected as part of a diagnostic study conducted at the author's institution on patients undergoing free flap reconstruction.

Time Range

January 2019 to June 2025

Freshness

Last updated 2026-05-11 17:23:43; freshness should be verified.