Data Sheet 1_Machine-learning prediction of 3- and 5-year mortality in lymph-node-positive

U.S. and Chinese patient data from the SEER database and Zibo Municipal Hospital, used to train and validate machine learning models for survival prediction. The dataset, created by Shaojun Yang and last updated in April 2026, contains 1,071 records from SEER and 198 for external validation. It was used to develop an online calculator for predicting 3- and 5-year overall and cancer-specific survival.

Use Cases

• Training survival prediction models based on clinical features like age, tumor size, and treatment history.
• Validating machine learning algorithms for clinical risk assessment using an external Chinese patient cohort.
• Investigating feature importance for cancer survival using SHAP analysis on factors like surgery, age, and tumor size.
• Developing clinical decision support tools for lymph-node-positive medullary thyroid carcinoma patients.

Strengths

• External validation performed on an independent Chinese cohort of 198 patients.
• Model performance metrics are reported, including AUC values up to 0.892 for 5-year overall survival prediction.
• Feature selection was performed using multiple methods (Cox, Boruta, RFE) and five machine learning algorithms were evaluated.

Limitations

• Row count for the primary dataset is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The data is stored in a DOCX file format, which is not ideal for direct machine learning use and may require extraction.

Provenance

Source

SEER database and Zibo Municipal Hospital.

Collection Method

Retrospective cohort study.

Freshness

Last updated 2026-04-14 04:14:59; freshness should be verified.

Geography

United States (SEER) and China (Zibo).