Age-Related Evolution Patterns in Online Handwriting.

Area of Science:

• Computational Linguistics
• Biometrics
• Gerontology

Background:

• Handwriting (HW) analysis is crucial for differentiating normal aging from neurodegenerative changes.
• Existing methods often assume a singular pattern of age-related HW evolution.

Purpose of the Study:

• To develop an original approach for online handwriting style characterization.
• To uncover handwriting style categories and their correlation with age.
• To identify distinct patterns of age-related handwriting changes.

Main Methods:

• A two-level clustering scheme was employed for handwriting style characterization.
• The first level created writer-independent word clusters from spatial-dynamic HW data.
• The second level generated a Bag of Prototype Words with interword stability for each writer, feeding into unsupervised learning.

Main Results:

• The study identified three major aging handwriting styles.
• One style was specific to aged individuals, while two were shared across age groups.
• Information theoretic measures quantified the age information gain from each clustering layer.

Conclusions:

• The proposed method effectively characterizes online handwriting styles and their age correlations.
• Handwriting aging is not monolithic; three distinct styles were discovered.
• This research advances the understanding of normal versus abnormal handwriting changes with age.