Predictive Model for Rituximab Response in Myasthenia Gravis: 117 Patient Records

117 patient records from a retrospective study between August 2019 and January 2025 were used to build a predictive model for minimal symptom expression after low-dose rituximab treatment. Author Chenlu Hou developed a logistic regression model presented as a nomogram, achieving an AUC of 0.777. The dataset was last updated on 2026-06-03 and is shared under a CC-BY-4.0 license.

Use Cases

• Validate predictive models for immunotherapy response based on clinical and biomarker variables.
• Train classifiers to stratify patients into high- and low-probability treatment response groups.
• Benchmark model performance using metrics like AUC, sensitivity, and specificity described in the study.

Strengths

• Model performance is quantified with an AUC of 0.777 and a Youden index-derived threshold of 0.534.
• The dataset is focused on a specific clinical question with 117 patients and five identified predictive variables.
• Includes internal validation via bootstrap method with 1000 replications for calibration and decision curve analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal and institutional bias inherent to the single-center, retrospective study design.

Provenance

Source

figshare

Collection Method

Retrospective review of patients who received RTX treatment at a research center.

Time Range

August 2019 to January 2025

Freshness

Last updated 2026-06-03 04:47:29; freshness should be verified.