Handwriting in Mild Cognitive Impairment: Reliability Assessment and Machine Learning-Based Screening

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Mild cognitive impairment (MCI) is a significant risk factor for dementia, necessitating early detection and monitoring.
• Current clinical tests for MCI have limitations, highlighting the need for objective, accessible tools.
• Quantitative handwriting analysis presents a promising noninvasive method for MCI assessment.

Purpose of the Study:

• To investigate the utility of quantitative handwriting analysis for noninvasive screening and monitoring of mild cognitive impairment (MCI).
• To assess the reliability and correlation of handwriting indicators with clinical assessments in MCI patients.
• To develop machine learning models for distinguishing MCI patients from healthy controls using handwriting data.

Main Methods:

• Utilized a sensorized ink pen to record handwriting data during daily life tasks (grocery list, free text) and a clinical dictation test (parole-non-parole).
• Computed 106 indicators related to time, fluency, force, and pen inclination from recorded data.
• Analyzed test-retest reliability, correlated indicators with clinical scores, and built machine learning classifiers for MCI detection.

Main Results:

• High reliability (93%) of indicators was observed for cursive handwriting, with moderate reliability (44%) for block letters.
• Better temporal handwriting performance correlated with preserved cognitive status and daily function.
• Machine learning models using free writing data achieved high accuracy (0.80-0.93) and F1-scores (0.81-0.92) in distinguishing MCI patients.

Conclusions:

• Ecological handwriting analysis is suitable for comprehensive MCI monitoring, from early screening to tracking disease progression.
• Sensorized pen technology provides objective, quantitative data for MCI assessment.
• This noninvasive approach holds potential for widespread clinical application in dementia prevention strategies.