Neurosyphilis Diagnosis Data for 1,648 Patients Across Four Medical Centers

1,648 suspected neurosyphilis cases from four medical centers in Guangzhou, Beijing, Xiamen, and Seattle were used to develop machine learning models for diagnosis. The dataset likely contains clinical features such as neurological symptoms, CSF protein, and CSF white blood cell count, which were identified as strong predictors. The research, authored by Yinbo Jiang and last updated in April 2026, compares models tailored to six international diagnostic guidelines.

Use Cases

• Training diagnostic machine learning models based on neurological symptoms and CSF assay results.
• Comparing the performance of different diagnostic guidelines (e.g., Random Forest, XGBoost) for neurosyphilis.
• Developing explainable AI tools for clinical decision support using SHAP analysis.
• Validating machine learning models across geographically diverse patient cohorts from China and the USA.

Strengths

• Includes 1,648 patient cases from four medical centers, providing a multicenter perspective.
• Models achieved excellent discrimination with AUC and PRAUC reported as greater than 0.90.
• Data supports comparison across six different international diagnostic guidelines.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying clinical data is unknown, which may limit suitability assessment.
• Data may reflect geographic bias inherent to the specific centers in China and the USA.

Provenance

Source

figshare, author Yinbo Jiang

Collection Method

Assembled from suspected neurosyphilis cases at four medical centers.

Freshness

Last updated 2026-04-17 02:21:10; freshness should be verified.

Geography

Guangzhou, Beijing, Xiamen (China) and Seattle (USA)