A machine learning model for predicting oligoclonal band positivity using routine cerebrospinal fluid and serum biochemical markers

Gozgoz H., Orhan O., Akan Konuk B., AKAN P.

AMERICAN JOURNAL OF CLINICAL PATHOLOGY, 2025 (SCI-Expanded, Scopus) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1093/ajcp/aqaf119
  • Dergi Adı: AMERICAN JOURNAL OF CLINICAL PATHOLOGY
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, BIOSIS, CINAHL, EMBASE
  • Anahtar Kelimeler: oligoclonal bands, cerebrospinal fluid, machine learning, multiple sclerosis, diagnostic prediction model
  • Dokuz Eylül Üniversitesi Adresli: Evet

Özet

Objective To develop and validate a machine learning model for predicting oligoclonal band (OCB) positivity using routine cerebrospinal fluid (CSF) and serum biochemical markers to improve the diagnostic accuracy and efficiency of assessing intrathecal immunoglobulin G (IgG) synthesis.Methods In this retrospective study (n = 1709), an ensemble model was developed using 8 refined CSF and serum parameters. Combining optimized CatBoost, XGBoost, and LightGBM classifiers, the model was trained and evaluated using a 2-phase workflow, including 5-fold cross-validation and validation on independent internal (n = 342) and external (n = 49) cohorts.Results The developed ensemble model achieved a receiver operating characteristic-area under the curve (ROC-AUC) of 0.902 on the internal test set, significantly outperforming the conventional IgG index (ROC-AUC, 0.795). At its optimal threshold, the model demonstrated an accuracy of 0.830, with a sensitivity of 0.714 and a specificity of 0.916. On the external validation cohort, it achieved 90% accuracy and 96% sensitivity.Conclusions A novel machine learning ensemble model accurately predicts OCB positivity using routine laboratory data and demonstrates superior performance compared with the IgG index. This approach represents a significant step in applying artificial intelligence in laboratory medicine, with the potential to enhance diagnostic efficiency. Prospective, multicenter validation is essential for broader clinical implementation.