CEU Electronic Theses and Dissertations, 2025
| Author | Rosic, Suncica |
|---|---|
| Title | Early Diagnosis of Parkinson's Disease through Machine Learning Analysis of Handwriting based on the PaHaW Dataset. Quantifying the economic burden of Parkinson???s Disease in Czechia. |
| Summary | This thesis investigates the predictive potential of handwriting in the early detection of Parkinson’s Disease (PD) (Study 1). Currently, PD diagnostics mostly involves clinical methods like imaging or Cerebrospinal fluid and serum tests. In recent years, we have observed a shift towards computer-aided diagnosis (CAD) based on behavioural data such as text, speech and handwriting. Despite contributions to the handwriting-based detection of PD, most of the literature has neglected the implications of disease severity. I use a PaHaW data set from Czechia, containing handwriting data from 75 participants, to identify features used to detect PD individuals in the early stages based on the UPDRS scale. I utilise the machine learning (ML) models Logistic Regression, Random Forest, SVC and XGBoost as binary classifiers to discriminate between PD and HC based on handwriting features. Results show that velocity parameters, pressure and number of changes in pressure constitute a diagnostic criterion for early detection of PD. In a longitudinal analysis of the economic implications of PD (Study 2), I estimate its economic burden in Czechia between 1996 and 2018, by utilising data from a variety of sources, including the IHME GBD database, Our World in Data and the Czech Statistical Office. Results show that PD prevalence among the working-age population represents a considerable economic burden, as productivity loss due to PD accounted for more than 0.02% of GDP in 2015. Those findings illustrate that handwriting-based CAD, statistics and ML are important tools for complementing clinical PD diagnostics to detect the disease earlier and reduce the economic burden. |
| Supervisor | Wittek, Mark |
| Department | Undergraduate Studies BA |
| Full text | https://www.etd.ceu.edu/2025/rosic_suncica.pdf |
Visit the CEU Library.
© 2007-2025, Central European University