Dimensions Matters In Decoding Infectious Diseases
For decades, the detection of infectious diseases has been limited by how data are analyzed. With the classic data analysis method, some truly infected individuals can be missed (‘false negative’ results), and a substantial number of infected and non-infected observations can be mixed – problems that result in misdiagnoses.
In a publication released today byPLOS ONE, University of New Mexico researchers Drs. Ariel L. Rivas and Douglas J. Perkins, along with an international network of collaborators, show a new way to address this double problem. Their approach explores the possibility that medical data might have different structural levels, which involve several dimensions. Instead of analyzing data on a flat format – uni-dimensional data as reported in any table; or bi-dimensional (2D) data, as any figure reported on a page or screen – the new method explores rotating three-dimensional (3D) data structures. This new method facilitates the expression ofanyfeature the data might have, which cannot be revealed by ‘flat’ formats.
When the 3D-based system is applied, humans, birds and cows display similar data patterns, regardless of whether they are infected by bacteria, virus, or parasites. The new model is robust, revealing patterns well-conserved throughout evolution.
“Our approach is similar to the way meaning or information emerges in general,” Rivas explains. “We need information, not just data. For example, when we look at any one letter, no information is obtained. However, when we combine letters, words emerge and, with them, meaning. If, in addition, we combine words; sentences emerge, which provide even more information. The higher the level or structure (words, sentences, paragraphs and beyond), the greater the information generated. We apply the same concept to biomedical data: the more dimensions and levels considered; the more interpretable and usable the information retrieved.”
Because this new model can detect false negatives, it drastically reduces the amount of mixed (infected and non-infected) observations, and to be implemented, it only requires leukocytes (the “white” cells found in fluids, such as blood and milk). This approach can be applied in virtually any medical setting in the world – one of the goals of UNM’s Center for Global Health.