Table 3_Exploratory machine learning analysis to characterize angioscopic features associa

A 2026 single-center pilot study by Satoru Takahashi analyzed 500 patients undergoing non-obstructive general angioscopy (NOGA). The dataset includes 56 patients with aortic dissection and 444 control patients with coronary artery disease, with aortic injuries classified into 13 types. A Random Forest model with LASSO feature selection and SHAP analysis was used to characterize features associated with aortic dissection.

Use Cases

• Training diagnostic classifiers for aortic dissection based on angioscopic features like intramural blood and puff sign.
• Interpreting feature importance in medical ML models using SHAP and Permutation Importance methods.
• Analyzing the relationship between different types of spontaneously ruptured aortic plaques and injuries (SRAPIs).
• Conducting network analysis to compare structural patterns between patient groups.
• Validating machine learning approaches for rare cardiovascular events in a clinical, observational setting.

Strengths

• Includes data from 500 patients (56 cases, 444 controls), providing a basis for comparative analysis.
• Applies multiple advanced ML techniques (Random Forest, LASSO, SHAP) for model development and interpretation.
• Classifies aortic injuries into 13 distinct, clinically relevant types for detailed analysis.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data is from a single-center, pilot study; the authors note validation in larger multicenter studies is required.

Provenance

Source

Satoru Takahashi via figshare.

Collection Method

Single-center, cross-sectional observational pilot study using invasive catheterization for Non-obstructive General Angioscopy (NOGA).

Freshness

Last updated 2026-05-07 05:44:24; freshness should be verified.