Development and validation of a diagnostic model to differentiate spinal tuberculosis from pyogenic spondylitis by...

Area of Science:

• Orthopedics and Spine Surgery
• Infectious Diseases
• Medical Diagnostics
• Machine Learning in Medicine

Background:

• Spinal tuberculosis (STB) and pyogenic spondylitis (PS) are distinct spinal infections with overlapping clinical presentations.
• Accurate differentiation between STB and PS is crucial for appropriate treatment and improved patient outcomes.
• Current diagnostic methods can be time-consuming and may not always provide definitive differentiation.

Purpose of the Study:

• To develop and validate a diagnostic model for differentiating between STB and PS.
• To identify key clinical and laboratory variables that distinguish STB from PS using machine learning.
• To create a reliable tool for healthcare practitioners to aid in the rapid and precise diagnosis of spinal infections.

Main Methods:

• Retrospective analysis of 387 confirmed cases of STB (n=241) and PS (n=146).
• Application of four machine learning algorithms (LASSO, logistic regression, random forest, SVM-RFE) to identify distinctive variables in a training group (n=271).
• Construction and validation of a diagnostic model using identified variables, assessed via ROC curves, calibration curves, and a validation group (n=116).

Main Results:

• Seven key variables were identified by machine learning algorithms to form the diagnostic model.
• The model achieved an Area Under the Curve (AUC) of 0.841 in the training group and 0.83 in the validation group.
• Significant differences were observed in platelet-to-neutrophil ratio (PNR), neutrophil-to-lymphocyte ratio (NLR), platelet volume distribution width (PDW), mean platelet volume (MPV), hemoglobin (HGB), and red blood cell (RBC) count between STB and PS patients.

Conclusions:

• The developed machine learning-based diagnostic model demonstrates high accuracy in differentiating between STB and PS.
• The model effectively utilizes readily available laboratory parameters to support clinical decision-making.
• This tool can facilitate faster and more precise diagnoses, leading to timely and appropriate management of spinal infections.