Machine-learning model may help in treatment decision-making in TED

A machine-learning model was able to predict the responsiveness to steroid treatment in patients with thyroid eye disease (TED) with relatively robust and reliable results in a small dataset, according to a study.

Factors found to possibly predict response to treatment were thyroid-stimulating immunoglobulin (TSI), extraocular muscle limitation, and low-density lipoprotein (LDL) cholesterol levels.

Using eXtreme Gradient Boosting (XGBoost), data from 89 patients with TED who received steroid treatment were retrospectively reviewed to determine their response to therapy.

The model had an accuracy of 0.861.

Less extraocular muscle limitation and high TSI levels were also found to be associated with a high risk of poor intravenous methylprednisolone treatment response, according to multivariate logistic regression analysis.

Reference
Park J, Kim J, Ryu D, et al. Factors related to steroid treatment responsiveness in thyroid eye disease patients and application of SHAP for feature analysis with XGBoost. Front Endocrinol (Lausanne). 2023;14:1079628. doi: 10.3389/fendo.2023.1079628. PMID: 36817584; PMCID: PMC9928572.