LASSO Model for Spinal Disease Diagnosis from 481 Patient Records

Liuzhou People’s Hospital provided data from 481 patients collected between June 2020 and December 2024. The dataset was used to develop and validate a machine learning model to differentiate spinal tuberculosis, pyogenic spondylitis, and endplate osteochondritis. The model is based on 12 core features and is accompanied by an online prediction tool.

Use Cases

• Training a diagnostic classifier for spinal tuberculosis based on the 12 core clinical and radiological features.
• Benchmarking machine learning algorithms for multi-class medical diagnosis using the reported performance metrics.
• Developing interpretable clinical prediction tools using SHAP analysis on the model's feature contributions.
• Creating a web-based clinical decision support calculator for the differential diagnosis of spinal conditions.

Strengths

• Includes data from 481 patients, providing a basis for model development.
• Model validation reports specific AUC values (0.838, 0.683, 0.897) for three disease classes.
• The study describes a clear methodology with a 7:3 train-validation split and uses 12 screened features.
• An open-access web-based prediction calculator was developed for clinical application.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying data is unknown, which may limit suitability assessment.
• Data is from a single hospital in Liuzhou, which may limit geographic generalizability.

Provenance

Source

figshare, author Tuo Liang.

Collection Method

Retrospective recruitment of patient cases from Liuzhou People’s Hospital.

Time Range

June 2020 to December 2024

Freshness

Last updated 2026-06-02 05:26:39; freshness should be verified.

Geography

Liuzhou, China (inferred from hospital location).