Wyniki wyszukiwania - BazTech

1

Feature projection k-NN classifier model for imbalanced and incomplete medical data

Porwik P., Orczyk T., Lewandowski M., Cholewa M.

Biocybernetics and Biomedical Engineering

|

2016

|

Vol. 36, no. 4

644--656

EN

Many datasets, especially various historical medical data are incomplete. Various qualities of data can significantly hamper medical diagnosis and are bottlenecks of medical support systems. Nowadays, such systems are often used in medical diagnosis. Even great number of data can be unsuitable when data is imbalanced, missing or corrupted. In some cases these troubles can be overcome by machine learning algorithms designed for predictive modeling. Proposed approach was tested on real medical data and some benchmarks dataset form UCI repository. The liver fibrosis disease from a medical point of view is difficult to treatment and has a significant social and economic impact. Stages of liver fibrosis are diagnosed by clinical observation and evaluations, coupled with a so-called METAVIR rating scale. However, these methods may be insufficient, especially in the recognition of phase of the disease. This paper describes a newly developed algorithm to non-invasive fibrosis stage recognition using machine learning methods – a classification model based on feature projection k-NN classifier. This solution allows extracting data characteristics from the historical data which may be incomplete and may contain imbalance (unequal) sets of patients. Proposed novel solution is based on peripheral blood analysis without using any specialized biomarkers, and can be successfully included to medical diagnosis support systems and might be a powerful tool for effective estimation of liver fibrosis stages.

2

Cancer classification based on gene expression data

Fujarewicz K., Rzeszowska-Wolny J.

Journal of Medical Informatics & Technologies

|

2000

|

Vol. 1

BI 23--27

EN

Proper classification of cancer is a crucial aspect in diagnosis and choosing optimal medical therapy. It has been suggested, in recent years, that classification process of cancer can be done using gene expression monitoring. Usefulness of this approach has increased due to the new technique of gene expression monitoring – using so called "expression chips". Recently in [1, 3] a heuristic method of cancer classification, called weighted voting (WV) method, based on gene expression levels has been proposed and tested on a set of samples of acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL). Here a more traditional approach to feature selection and classification is presented and tested on the same data set. Feature selection is performed using modified Sebestyen criterion and classification is done using linear classifying function trained by modified perception algorithm. Obtained results are better than results of the WV method. In cross-validation of initial set all 38 samples were classified correctly (WV – 1 incorrect) and only one sample from independent set was classified incorrectly (WV – 2 incorrect).

3

Some properties of solutions for two-dimensional choice problems reconsidered

Lahiri S.

Control and Cybernetics

|

1999

|

Vol. 28, no 1

75-87

EN

In this paper, we take up the problem of axiomatically characterizing what we have referred to in the paper as the additive choice function on the classical domain for choice problems. Apart from an impossibility result for the additive choice function, there is an axiomatic characterization, which as a by-product provides a counterexample to a conjecture for the egalitarian choice function. In an appendix, we provide a proof of an axiomatic characterization of the egalitarian choice function using a superadditivity axiom. In this paper, we also provide proofs of axiomatic characterizations of the family of non-symmetric Nash choice functions and the family of weighted hierarchies of choice functions. Our conclusion is that earlier axiomatizations are essentially preserved on the classical domain for choice problems. The proofs are significant in being non-trivial and very dissimilar to existing proofs for other domains.