How word semantics and phonology affect handwriting of Alzheimer’s patients - a machine learning based analysis

Aug 4, 2026

Upload by Ivy

Using kinematic properties of handwriting to support diagnosis of neurodegenerative disease remains challenging, and non-invasive detection combined with machine learning offers promising progress. The study examines how word semantics and phonology influence handwriting in individuals affected by Alzheimer’s disease. Six copying tasks were used with words grouped into regular, non-regular, and non-word categories. Four widely used classifiers and feature selection were applied to extract distinctive features per word type. Results show non-regular words require more features yet reach accuracy close to 90%.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Abstract

Introduction

Background and motivation

Prior handwriting studies

Role of machine learning and feature selection

How word semantics and phonology affect handwriting of Alzheimer’s patients: a machine learning based analysis Nicole D. Ciliaa,b , Claudio De Stefanoc , Francesco Fontanellac,1 , Sabato Marco Siniscalchia a Department of Computer Engineering, University of Enna ”Kore”, Italy b Institute for Computing and Information Sciences, Radboud University Nijmegen, The Netherlands c Department of Electrical and Information Engineering Mathematics, University of Cassino and Southern Lazio, Italy Abstract Using kinematic properties of handwriting to support the diagnosis of neurodegenerative disease is a real challenge: non-invasive detection techniques combined with machine learning approaches promise big steps forward in this research field. In literature, the tasks proposed focused on different cognitive skills to elicitate handwriting movements. In particular, the meaning and phonology of words to copy can compromise writing fluency. In this paper, we investigated how word semantics and phonology affect the handwriting of people affected by Alzheimer’s disease. To this aim, we used the data from six handwriting tasks, each requiring copying a word belonging to one of the following categories: regular (have a predictable phoneme-grapheme correspondence, e.g., cat), non-regular (have atypical phoneme-grapheme correspondence, e.g., laugh), and non-word (non-meaningful pronounceable letter strings that conform to phoneme-grapheme conversion rules) . We analyzed the data using a machine learning approach by implementing four well-known and widely-used classifiers and feature selection. The experimental results showed that the featureselection allowed us to derive a different set of highly distinctive features for each word type. Furthermore, non-regular words needed, on average, more features but achieved excellent classification performance: the best result was obtained on a non-regular, reaching an accuracy close to 90% . 1. Introduction In recent decades, human movement motor activities have been investigated with increasing frequency with the aim of discovering the underlying cognitive processes. Researchers have found that cognitive processing and the brain motor system are not functionally independent: an individual movement is the end result of a cognitive process. They also found that the relationship between the two systems is much more complex than previously imagined [1] . Furthermore, more recent findings have shown that alterations in motor activities can bea prodromal sign of neurodegenerative diseases. For example, people affected by Alzheimer’s Disease (AD) exhibit alterationsin spatial organization and poor control of fine movements. In this context, the analysis of the alterations in handwriting can be very useful since handwriting is the result of complex interac- make an early diagnosis, which remains challenging. It is worth noting that early diagnosis is particularly important for AD: the currently available treatments are much more effective in slowing the course of this incurable disease when started early. Even more importantly, also disease-modifying pharmacological treatments that will be available in the near future have shown to be effective only if started in the early stages of AD [4] . Until a few years ago, most of the studies analyzing the effect of AD on handwriting were conducted by physicians and psychologists. They typically based their analysis on data collected from a few dozen participants while performing a few handwriting tasks [5] . Furthermore, those studies typically used statistics-based approaches, e.g., the Pearson correlation coefficient, focused on the relationship between the disease and each arXiv :2307 .04762v 1 [ cs .CL] 6 Jul 2023 tions between bio-mechanical parts (arm, wrist, hand, etc.) and brain areas devoted to the control and memorization of the elementary motor sequences used to produce handwritten traces [2] . For example, in the clinical course of AD, dy

FAQs

What is the main objective of the study?

To investigate how word semantics and phonology affect the handwriting of people with Alzheimer’s disease, using machine learning on handwriting kinematics.

How are the words in the tasks categorized?

The study uses six copying tasks where the stimuli belong to regular words, non-regular words, or non-words.

What was the key outcome of applying feature selection and classifiers?

Feature selection produced different highly distinctive feature sets for each word type, and non-regular words achieved the best classification performance with accuracy near 90%.

Aug 4, 2026

Upload by Ivy