A Two-Stage Tremor Detection Algorithm for Wearable Inertial Sensors During Normal Daily Activities

Area of Science:

• Neurology
• Biomedical Engineering
• Wearable Technology

Background:

• Continuous tremor monitoring via wearable sensors during daily activities is challenging due to overlapping movement frequencies and wide amplitude variations.
• Distinguishing pathological tremor from normal physiological movements in real-world settings requires sophisticated algorithms.

Purpose of the Study:

• To introduce and validate a novel two-stage algorithm for improved tremor detection using wearable wrist sensors.
• To enhance the accuracy of continuous tremor monitoring during normal daily activities.

Main Methods:

• A two-stage algorithm was developed, incorporating prior knowledge of individual tremor frequency ranges.
• The algorithm was validated using continuous wrist sensor recordings from Parkinson's disease patients and control subjects.
• Performance was assessed based on face validity, false positive rates in controls, and correlation with clinical tremor assessments.

Main Results:

• The algorithm demonstrated improved discrimination between tremor and other daily activities.
• A low false positive rate (<1.1%) was observed in control subjects, indicating high specificity.
• The algorithm showed good correspondence with the MDS-UPDRS rest tremor constancy measure (ρ = 0.54).

Conclusions:

• The novel two-stage algorithm effectively distinguishes tremor from normal movements during daily activities.
• This approach offers a promising method for accurate, continuous tremor monitoring in clinical populations.
• The validated algorithm supports the use of wearable sensors for objective tremor assessment outside clinical settings.