Cell Biochemistry Martinsried

Cytomics

G.Valet



Cytomics as multimolecular cytometric analysis of cells and cell systems (cytomes) in combination with an exhaustive bioinformatic knowledge extraction of the analysis results,
access a maximum of information on the apparent molecular cell phenotype as it results from cell genotype and exposure.

Molecular cell phenotypes in the naturally existing cellular and cell population heterogeneity of disease affected body cytomes contain information on the future development (prediction) as well as on the present status (diagnosis) of diseases, since diseases are caused by molecular changes in cell systems or organs. With the concept, from cells to patient, the analysis of the full heterogeneity of cellular data opens the way for therapy dependent individualised disease course predictions for the general practice of medicine. Predictions provide a therapeutic lead time. Early preventive therapies can for example try to prevent irreversible tissue damage. It may also be possible to achieve disease retardation or prevention in certain situations like for potential asthma patients by the early recognition of a beginning sensibilisation phase. The immediate sanitation of patients environment may then postpone or prevent disease declaration.

Data classifications are presently considered predictive for individual patients at predictive values >95% for each classified disease category of the learning set while they are prognostic at values <95%. The effort will be to elevate this level to >99% through the search for more efficiently discriminating molecular data patterns.

General concept
for predictive medicine by cytomics:
a.) multiparametric cytometric determination of cell constituents or cell functions in disease associated cytomes
b.) analysis (1, 2) of all measured numeric parameters for all cell populations that is in practice for >95% of the collected cells
c.) data pattern classification of this entire information against patient's future disease course during the learning phase by exhaustive knowledge extraction
d.) classification of the embedded test set of patient data, measured under the same conditions as the learning set but remaining unknown to the learning process. Typically, every 5th or 10th patient is assigned to the test set prior to the learning phase to exclude classification biases.
e.) prospective classification of data collected from subsequent new patients during the clinical evaluation phase



2004 G.Valet
Max-Planck-Institut für Biochemie, Am Klopferspitz 18a, D-82152 Martinsried, Germany, Tel: +49/89/8578-2518, -2525, Fax: +49/89/8578-2563, INTERNET: http://www.biochem.mpg.de/valet/cellbio.html
Last Update: Apr.10,2004