Despite the challenge of distinguishing tics from extra movements, machine learning technology could potentially help researchers with reducing time spent analyzing video recordings of patients with tic disorders.
According to a recent study presented at the 2023 International Congress of Parkinson’s Disease and Movement Disorders, held August 27-31, in Copenhagen, Denmark, findings showed that machine learning could effectively detect and distinguish tics from extra movements among patients with tic disorders.1 These results suggest that the machine learning algorithm could be developed into a clinically applicable tool to be used for enhancing classification accuracy in the recognition of tics in patients.
In a dataset of 63 videos of patients with tic disorders, use of the Random Forest classifier resulted in an 83% accuracy rate in identifying patients and healthy controls with tics and extra movements. Notably, paired samples t-tests displayed significant differences between the 2 groups of patients in all features derived from the tic predictions.
“The frequency and characteristic cluster aggregation of tics are key determinants of tic severity. Wearable sensors recording tics in patients’ natural environment are currently under exploration, but the anatomical distribution and diverse phenomenology of tics hinder the routine clinical applicability of these sensors. Tic frequency and phenomenology are also routinely assessed using video recordings usually obtained in a clinical setting, a methodology often used also in clinical trials,” Davide Martino, MD, PhD, professor of neurology at the University of Calgary, said about the study in a statement.1
In this study, investigators aimed to develop a machine learning tool to distinguish between facial/head tics among patients with Tourette syndrome and spontaneous movements and healthy controls. For identifying tic movements, a Random Forest classifier was used since it recognized facial landmarks as input and defined tic seconds as those with tics of equal or greater severity than a predefined threshold. Thus, the trained classifier was used to predict the presence of tics in patients as well as any extra movements in healthy controls.
The predictions were used to calculate features including the number of tics per minute, maximum duration of a continuous nontic segment, maximum duration of a continuous tic, average duration of ticfree segments, number of changes from tic to nontic segments and vice versa per minute, average size of a tic-cluster, and the number of clusters per minute. The features were then combined into a single tic detection score using logistic regression. The parameters of the model were collected by testing a dataset of 124 videos of patients with tic disorders and 162 videos of healthy controls. Additionally, to explore the accuracy of the score in classifying patients and healthy controls, a test dataset of 50 videos of patients and 50 videos of healthy controls was utilized.
“Rating these recordings is time- and energy-consuming. This study applies machine learning to train an algorithm that classifies tics from non-tic extra movements and measures several parameters detailing the temporal distribution of tics, ultimately combining these into a single tic detection score. The study reports a very good classification accuracy of the algorithm (83%), although the composition and accuracy of the tic detection score is still in progress,” Martino said in a statement.1
Researchers noted that their next step is to fine-tune the tic detection score in order to improve classification accuracy.2 They are aiming to analyze the significance of each feature to determine which characteristics are most helpful in differentiating between the 2 groups and also seeing if the algorithm is helpful in distinguishing between tics and functional movements.
“An algorithm that measures frequency and clustering of tics from video recordings has strong translational value in routine clinical practice and clinical research, as it would likely optimize reliability and efficiency of these measurements,” Martino said in a statement.1 “Although limited to facial/head tics, the same approach can be extended to other body regions and phonic tics. It is also important to point out that video recording-based measures will inevitably still need to be integrated with other domains of tic severity, e.g., interference with daily routines and functional impact, in order to achieve a truly comprehensive assessment of tics.”