Interpretation of Change in Novel Digital Measures: A Statistical Review and Tutorial

Area of Science:

• Digital health technology
• Clinical outcome assessment
• Statistical methodology

Background:

• Novel clinical measures from digital health tools require thresholds for interpreting change over time.
• Establishing thresholds, like the minimal clinically important difference, is vital for clinical validation.
• These thresholds aid in understanding the clinical relevance of treatment effects.

Purpose of the Study:

• To present the theoretical background for interpretative thresholds in digital clinical measures.
• To review methods for estimating these thresholds, including anchor-based approaches.
• To illustrate threshold estimation using real-world clinical trial data.

Main Methods:

• Application of established methods for patient-reported outcomes to digital clinical measures.
• Distinguishing between thresholds based on perceived importance versus measurement error.
• Considering thresholds for both group- and individual-level interpretations.
• Utilizing anchor-based approaches for threshold estimation.

Main Results:

• Demonstrated applicability of existing threshold derivation methods to digital clinical measures.
• Provided a framework for understanding different types of interpretative thresholds.
• Illustrated threshold estimation with cough frequency data from a wearable device.

Conclusions:

• Statistical methodologies can effectively estimate thresholds for interpreting change in digital clinical measures.
• Thresholds are essential for the clinical validation and meaningful interpretation of digital health data.
• The presented methods facilitate a better understanding of treatment effects in clinical trials.