EXMD2382 Lecture Notes - Lecture 1: Dual-Energy X-Ray Absorptiometry, Repeated Measures Design, False Positives And False Negatives
Lecture 1 – Technical Skill Accuracy
Measurement Process
• Test – what you do (protocol)
• Measure – how you do it (follow protocol)
Accuracy – how close you are to the true value
Calibration – adjusting a device and determine its accuracy by comparing it to a standard
Verification – checking (not adjusting) how accurate the equipment is against the standard
Standards
• Gold standard – the most accurate test (e.g., VO2max done with gas analysers; often referred
to as a cardiopulmonary exercise test [CPET])
• Reference method – a thoroughly investigated test in which the precision and accuracy is
known (e.g., DEXA, reference material)
Precision (reliability) – how similar are repeated measures
• Test-retest reliability
- Intra-tester reliability (same tester)
- Inter-tester reliability (different testers)
Factors affecting precision
- The measurement device
- Within-individual variation (e.g., HR, the person might be less nervous with a different
tester)
Calculating Accuracy
1. Percent error – express the difference between the measured value and the true value as a
percentage
- HR monitor = 100 bpm
- Manual method = 90 bpm
- Percent error = 10/100 = 10%
Document Summary
Measurement process: test what you do (protocol, measure how you do it (follow protocol) Accuracy how close you are to the true value. Calibration adjusting a device and determine its accuracy by comparing it to a standard. Verification checking (not adjusting) how accurate the equipment is against the standard. Precision (reliability) how similar are repeated measures: test-retest reliability. Within-individual variation (e. g. , hr, the person might be less nervous with a different tester) Calculating accuracy: percent error express the difference between the measured value and the true value as a percentage. Percent error = 10/100 = 10: correlation coefficient (between 0 and 1) correlates two measures to get an r-value (the closer to 1, the more accurate the measure) Bland-altman plot graphical (visual) method of comparing the difference between two measurements (if results are in 95% limits of agreement, then it is an accurate measure)