The following is a summary of “Parkinson’s disease tremor prediction using EEG data analysis-A preliminary and feasibility study,” published in the November 2023 issue of Neurology by Farashi et al.

Predicting tremor initiation in Parkinson’s disease (PD) could optimize deep brain stimulation (DBS) as a complementary treatment for medication-resistant tremors.

Researchers performed a retrospective study to develop a novel method for predicting resting tremors based on EEG time-series analysis.

They introduced a modified algorithm for detecting tremor onset from accelerometer data. Additionally, they presented a machine learning approach for predicting PD hand tremors using EEG time series. The classifier, trained on essential EEG features identified through statistical analyses, distinguishes pre-tremor conditions.

The results showed statistical analyses with post-hoc tests, form factor, and statistical features were the most discriminative. Additionally, a precise tremor prediction based on EEG data could be achieved with minimal EEG channels (F3, F7, P4, CP2, FC6, and C4) and EEG bands (Delta and Gamma). Utilizing the chosen feature set, a KNN classifier achieved optimal pre-tremor prediction performance, achieving an accuracy of 73.67%.

Investigators concluded that the EEG time-series analysis is feasible for predicting PD hand tremors, necessitating further research with more extensive data and diverse brain dynamics for clinical applications.

Source: bmcneurol.biomedcentral.com/articles/10.1186/s12883-023-03468-0