Authors
Jheremy S Reyes, Alexandros Bouras, Constantinos G Hadjipanayis, L Dade Lunsford, Ajay Niranjan
Published in
Acta neurologica Belgica. Nov 06, 2026. Epub Nov 06, 2026.
Abstract
Gamma Knife (GK) thalamotomy is a minimally invasive treatment option for medically refractory essential tremor (ET), but clinical improvement is typically delayed for several months. Machine learning may enable individualized prediction of treatment response using routinely available data at the time of treatment.Structured abstract.
We retrospectively analyzed 163 GK thalamotomy procedures performed in 139 patients with ET. A random forest model was developed using demographic, clinical, baseline Fahn-Tolosa-Marín (FTM), and radiosurgical planning variables available before or at treatment. The primary outcome was percentage improvement in a four-domain composite FTM score. Validation included patient-level 80/20 hold-out testing and repeated 5-fold cross-validation. Performance was assessed using R², mean absolute error (MAE), and root mean squared error (RMSE).
On hold-out validation, the model achieved an R² of 0.79 and an MAE of 16 percentage points. Repeated 5-fold cross-validation demonstrated a mean R² of 0.72 (95% CI, 0.64-0.79), mean MAE of 15.8 percentage points (95% CI, 14.6-17.1), and mean RMSE of 20.4 percentage points (95% CI, 18.7-22.2). Exploratory prediction of time to clinical benefit yielded an R² of 0.31, MAE of 1.78 months, and RMSE of 2.34 months.
Machine learning-based prediction of tremor improvement using data available on the day of GK thalamotomy is feasible and may support more personalized patient counseling and outcome forecasting in functional radiosurgery.
PMID:
42993427
Bibliographic data and abstract were imported from PubMed on 07 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 8
- Comments 0