The inter-assay precision was ≤12.1 % CV and the inter-assay relative accuracy was ≤10.9 % absolute bias, bringing the total error of the assay to ≤23%%.
Numerical accuracy was assessed by absolute relative difference (ARD) and International Organization for Standardization (ISO) 15197:2013 15/15% limits, and clinical accuracy was assessed by Clarke error grid analysis.
Utilizing the Multi-Agent Collaborative Learning Architecture, 84.5% absolute accuracy was obtained, an improvement of about 6.5% over the best results achieved by the Kansas Geological Survey with the same data set.
The absolute accuracy was defined by the root mean square error (RMSE) between the coordinates measured by the camera system and by engineering surveying (micro-triangulation).
Comparisons of field measured ρs with those derived from HELM corrected SPOT imagery, covering Helsinki, Finland, and Taita Hills, Kenya, indicated HELM-1 RFR absolute accuracy was ±0.02ρs in the visible and near infrared (VIS/NIR) bands and ±0.03ρs in the shortwave infrared (SWIR), whilst HELM-2 performance was ±0.03ρs in the VIS/NIR and ±0.04ρs in the SWIR.
Determination of absolute accuracy was based on the following statistics of the measurement error.
The threshold oxygen tension was reported in this study with the caveat that its absolute accuracy was dependent on assumptions made for the mathematical model and measurements of key parameters such as ρcell and Qcell.
Diagnostic accuracy was good for detection of absolute EELV changes above 200 mL (AUC = 0.79).
For the purpose of this survey, relative accuracy was of more concern than absolute accuracy, so Global Navigation Satellite Systems (GNSS) observations were chosen.
Diagnosis accuracy was fair for detection of absolute EELV changes above 200 mL (AUC 0.79 (CI 95% 0.70 to 0.89)), and good for detection of absolute EELV changes above 300 mL (AUC 0.89 (CI 95% 0.83 to 0.95)).
At first, the latency data might suggest participants did not fully engage with the secondary task in the low interactivity condition, yet accuracy was down by 30% and absolute calculation errors were four times as large with articulatory suppression in the low interactivity condition (see Fig. 1).
