Unbiased characterization of tremor disorders using Deep Learning for time series analysis
05.06.2023Massive time series feature extraction is a powerful tool to examine oscillating biological signals (Fulcher and Jones, 2017) such as tremor disorders. Tremor is defined as the involuntary, rhythmical, sinusoidal movement of a limb, such as for example in Parkinson´s disease or essential tremor. The exact phenomenological study of tremor movements has been shown to predict the response to non-invasive stimulation of an individual tremor (Schreglmann et al., 2021).
This study is aiming to extend an existing analytical pipeline examining higher-dimensional mathematical features extracted from accelerometer time-signals recorded over the tremulous limb. In addition, we intend to utilize deep learning approaches, such as long-short term networks (LSTMs) and Convolutional Neural Networks (CNNs). The pipeline will be used to differentiate tremor causes, response to treatment and explore tremor disorders in a more mechanistic way.
Fulcher BD, Jones NS. hctsa: A Computational Framework for Automated Time-Series Phenotyping Using Massive Feature Extraction. Cell systems 2017; 5: 527-531.e3.
Schreglmann SR, Wang D, Peach RL, Li J, Zhang X, Latorre A, et al. Non-invasive suppression of essential tremor via phase-locked disruption of its temporal coherence. Nat Commun 2021; 12: 36
Requirements
Experience programming with Python (or Matlab)
(Beneficial) Experience with Pytorch or Tensorflow
(Beneficial) Interest in data-driven neuroscience.
Betreuer: Anna Krause
