Measurement Reliability in the Early States of Biomarker Development
Walter S. Liggett Jr, Peter E. Barker, J Semmes
Analytical instruments with functional responses such as SELDI-TOF mass spectra offer a basis for biomarker development. This paper describes an approach to improving measurement reliability, that is, to improving the consistency of the instrument response, through assessment of sources of variation. The approach is suitable for instruments with functional responses and can therefore be applied even if clinical interpretation of the response has not yet been fully specified. The approach involves an experiment in which measurement of a reference material is replicated with a source of variation sometimes held fixed and sometimes not. The experimental results are interpreted by means of functional principal components analysis. In our illustration, the functional responses are SELDI-TOF mass spectra, and the source of variation is the difference between protein biochips. Among other things, the experiment shows that the measurement-to-measurement deviations in the heights of spectral peaks have complicated statistical dependencies. The chip-to-chip variation contributes to these deviations but not in an overwhelming way. The paper concludes with a discussion of the need for addition of metrological studies such as the one presented to the case-control studies usually envisioned in biomarker development.
functional data analysis, guage R & R, mass spectrum, measurement probability, proteomics, statistics
Liggett Jr, W.
, Barker, P.
and Semmes, J.
Measurement Reliability in the Early States of Biomarker Development, Disease Markers
(Accessed February 24, 2024)