Laboratory Data Analysis System: Section III—Multivariate Normality

TLDR

Multivariate analysis relies on sampling normal populations, a transformation to achieve Gaussian uniformity, and prior data‑analysis experience, and is positioned as a phase‑two component of a Laboratory Data Analysis System. The paper introduces a new laboratory data interpretation method called multivariate analysis. The method screens raw data to detect significant abnormalities and flags the system for further investigation. Illustrations of two‑ and three‑dimensional models on real data demonstrate that multivariate analysis yields more specific interpretations, resolves the abnormal‑normal issue, and offers a mathematical basis for advanced data‑processing techniques that condense large information volumes for clinical decision making.

Abstract

A new form of laboratory data interpretation, called “multivariate analysis,” is described. This process is feasible because of a sampling of normal people, a transformation scheme designed to create variable uniformity around a standard Gaussian distribution, and a practical experience of data analysis which shows the technic to be functioning. Examples are shown for two- and three-space models as applied to actual subject data. Multivariate analysis allows more specificity of interpretation, eliminates the problem of the “abnormal-normal,” and provides a mathematical foundation for an advanced data processing technic which can summarize vast amounts of information in a convenient mode for clinical decision making. The process is described as a phase two element in a Laboratory Data Analysis System . In this situation, multivariate analysis serves to screen the raw data to detect any significant abnormality and flag the system for further investigation.