LASSO Model for Differentiating Spinal Diseases: 481 Patient Cases from 2020-2024

A dataset of 481 patient records used to develop a machine learning model for differentiating spinal tuberculosis, pyogenic spondylitis, and endplate osteochondritis. The data was retrospectively collected from Liuzhou People’s Hospital between June 2020 and December 2024. The study constructed a LASSO regression model based on 12 core features and validated it with an internal cohort.

Use Cases

• Training a multi-class classifier for spinal disease diagnosis based on clinical and radiological features.
• Benchmarking machine learning algorithms (six were evaluated) for medical diagnostic tasks.
• Conducting feature importance analysis using SHAP to interpret model predictions for clinical decision-making.
• Developing a clinical prediction calculator, as referenced by the web-based tool created in the study.

Strengths

• Includes 481 patient records with confirmed diagnoses across three spinal conditions.
• Model validation used a defined 7:3 train-test split, with 338 and 143 patients respectively.
• The LASSO model achieved specific AUC scores (e.g., 0.897 for endplate osteochondritis) in the validation cohort.
• Features model interpretability through SHAP analysis, as described in the methods.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the dataset file is unknown, which may limit suitability assessment.
• Data reflects a single-hospital source (Liuzhou People’s Hospital), which may limit generalizability.

Provenance

Source

Liuzhou People’s Hospital, author Tuo Liang.

Collection Method

Retrospectively recruited patient cases.

Time Range

June 2020 to December 2024.

Freshness

Last updated 2026-06-02 05:26:40; freshness should be verified.

Geography

Likely China, based on hospital name.