Quantifying Nocturnal Scratch in Atopic Dermatitis: A Machine Learning Approach Using Digital Wrist Actigraphy

Area of Science:

• Biomedical Engineering
• Dermatology
• Digital Health

Background:

• Nocturnal scratching significantly impacts the quality of life for atopic dermatitis (AD) patients.
• Current methods for measuring scratch, primarily patient-reported outcomes (PROs), lack objectivity and sensitivity.
• Digital health technologies (DHTs) offer potential for objective behavioral monitoring.

Purpose of the Study:

• To develop and validate a machine learning algorithm for objectively quantifying nocturnal scratching events using wrist-worn actigraphy.
• To improve the assessment of disease progression, treatment effectiveness, and quality of life in AD patients.

Main Methods:

• Development and comparison of several machine learning models.
• Utilizing wrist-worn actigraphy data to capture nocturnal scratching behaviors.
• Validation of the algorithm on data from seven subjects in an inpatient setting.

Main Results:

• The best-performing machine learning model achieved an F1 score of 0.45 on the test set.
• The model demonstrated a precision of 0.44 and a recall of 0.46 for scratch detection.
• Preliminary results show promise for objective scratch quantification.

Conclusions:

• An automatic scratch detection algorithm using actigraphy has the potential to objectively assess sleep quality and disease state in AD patients.
• Further validation with a larger subject pool is required.
• This advancement could refine therapeutic strategies for AD management.