Using a prediction approach to assess agreement between two continuous measurements

Area of Science:

• Medical device evaluation
• Statistical methods in medicine
• Biostatistics

Background:

• Assessing agreement between medical devices is crucial for approving new technologies.
• Current methods may not adequately predict future performance consistency.
• Reliable agreement assessment ensures patient safety and effective clinical practice.

Purpose of the Study:

• To present a statistical method using prediction intervals for evaluating agreement between two measurement devices.
• To provide a framework for estimating the confidence in future differences between devices.
• To illustrate the application of prediction intervals in a practical medical context.

Main Methods:

• The study employs prediction intervals to estimate the range of future differences between paired measurements.
• A statistical approach is detailed for calculating and interpreting these intervals.
• The method is demonstrated using real-world data from peak expiratory flow measurements.

Main Results:

• Prediction intervals offer a quantifiable measure of confidence for future device agreement.
• The proposed method allows for the estimation of acceptable performance margins.
• The example demonstrates the practical utility of prediction intervals in device assessment.

Conclusions:

• Prediction intervals provide a robust statistical tool for assessing medical device agreement.
• This method enhances the reliability of approving new devices based on performance consistency.
• The approach is valuable for ensuring the continued accuracy and comparability of medical measurements.