Evaluation of Fractional Calculus and Delta Parameters in Prodromal Diagnosis of Dementia with Lewy Bodies utilizing Online Handwriting

Evaluation of Fractional Calculus and Delta Parameters in Prodromal Diagnosis of Dementia with Lewy Bodies utilizing Online Handwriting

Paper in proceedings (conference paper)

This study evaluates the effect of fractional order derivatives (FD) and delta parameters in the prodromal diagnosis of dementia with Lewy bodies (DLB) utilizing online handwriting analysis. With DLB being the second most prevalent neurodegenerative dementia, early detection is critical for timely intervention. Leveraging advanced mathematical models, we explored the potential of FD-based and delta-based kinematic handwriting features compared to baseline. The analysis included 45 participants at high risk of developing DLB and 29 healthy controls who performed the Archimedean spiral task. Our findings reveal that FD-based kinematic features, mainly derived from the horizontal velocity at low alpha levels, are significantly discriminative. Moreover, the binary classification model trained with FD-based features achieved a balanced accuracy BACC=0.75. The study emphasizes the relevance of advanced kinematic parametrization in neurodegenerative disease diagnostics, presenting novel features as promising tools for DLB screening.

This study evaluates the effect of fractional order derivatives (FD) and delta parameters in the prodromal diagnosis of dementia with Lewy bodies (DLB) utilizing online handwriting analysis. With DLB being the second most prevalent neurodegenerative dementia, early detection is critical for timely intervention. Leveraging advanced mathematical models, we explored the potential of FD-based and delta-based kinematic handwriting features compared to baseline. The analysis included 45 participants at high risk of developing DLB and 29 healthy controls who performed the Archimedean spiral task. Our findings reveal that FD-based kinematic features, mainly derived from the horizontal velocity at low alpha levels, are significantly discriminative. Moreover, the binary classification model trained with FD-based features achieved a balanced accuracy BACC=0.75. The study emphasizes the relevance of advanced kinematic parametrization in neurodegenerative disease diagnostics, presenting novel features as promising tools for DLB screening.

biomedical signal processing, feature extraction, fractional calculus, delta parameters, dementia with Lewy bodies, online handwriting, prodromal diagnosis

biomedical signal processing, feature extraction, fractional calculus, delta parameters, dementia with Lewy bodies, online handwriting, prodromal diagnosis

MUCHA, J.; GAVENČIAK, M.; MEKYSKA, J.; FAÚNDEZ ZANUY, M.; BRABENEC, L.; REKTOROVÁ, I.

2025

26.08.2024

EURASIP

Lyon, France

978-9-4645-9361-7

2024 32nd European Signal Processing Conference (EUSIPCO)

32

1731

1735

5

https://ieeexplore.ieee.org/document/10715381