A comparison of methods to generate adaptive reference ranges in longitudinal monitoring

In a clinical setting, biomarkers are typically measured and evaluated as biological indicators of physiological state. Population based reference ranges, known ‘static’ or ‘normal’ often used tool to classify biomarker value for an individual typical atypical. However, these ranges may not be informative particular when considering changes in over time since each observation is assessed isolation against the same limits. To allow early detection unusual changes, adaptation static required that incorporates within-individual variability arising from longitudinal monitoring addition between-individual variability. overcome this issue, methods generating individualised proposed within Bayesian framework which adapts successively whenever new measurement recorded individual. This approach also allows differ individual, compared other less flexible approaches. usually comes with high computational cost, especially individuals large number observations, diminishes its applicability. difficulty suggests approximation required. Thus, adaptive by use time-efficient approximate Expectation-Maximisation (EM) algorithm will presented relies only on few sufficient statistics at level.