RESUMEN
BACKGROUND: Patients with chronic kidney disease stage 5 (CKD5) are predisposed to vascular calcification (VC), but the combined effect of factors associated with VC was sparsely investigated. We applied the relaxed linear separability (RLS) feature selection model to identify features that concomitantly associate with VC in CKD5 patients. METHODS: Epigastric arteries collected during surgery from living donor kidney transplant recipients were examined to score the histological extent of medial VC. Sixty-two phenotypic features in 152 patients were entered into RLS model to differentiate between no-minimal VC (n = 93; score 0-1) and moderate-extensive VC (n = 59; score 2-3). The subset of features associated with VC was selected on the basis of cross-validation procedure. The strength of association of the selected features with VC was expressed by the absolute value of 'RLS factor'. RESULTS: Among 62 features, a subset of 17 features provided optimal prediction of VC with 89% of patients correctly classified into their groups. The 17 features included traditional risk factors (diabetes, age, cholesterol, BMI and male sex) and markers of bone metabolism, endothelial function, metabolites, serum antibodies and mitochondrial-derived peptide. Positive RLS factors range from 1.26 to 4.05 indicating features associated with increased risk of VC, and negative RLS factors range from -0.95 to -1.83 indicating features associated with reduced risk of VC. CONCLUSION: The RLS model identified 17 features including novel biomarkers and traditional risk factors that together concomitantly associated with medial VC. These results may inform further investigations of factors promoting VC in CKD5 patients.
Asunto(s)
Insuficiencia Renal Crónica/patología , Calcificación Vascular/patología , Adulto , Factores de Edad , Anciano , Índice de Masa Corporal , Colesterol/sangre , Complicaciones de la Diabetes/patología , Femenino , Humanos , Masculino , Persona de Mediana Edad , Fenotipo , Insuficiencia Renal Crónica/complicaciones , Factores de Riesgo , Índice de Severidad de la Enfermedad , Factores Sexuales , Calcificación Vascular/etiología , Adulto JovenRESUMEN
OBJECTIVES: To improve the medical diagnosis support rules based on comparisons of diagnosed patients with similar cases (precedents) archived in a clinical database. The case-based reasoning (CBR) or the nearest neighbors (K-N) classifications, which operate on referencing (learning) data sets, belong to this scheme. METHODS: Inducing similarity measure through special linear transformations of the referencing sets aimed at the best separation of these sets. Designing separable transformations can be based on dipolar models and minimization of the convex and piecewise linear (CPL) criterion functions in accordance with the basis exchange algorithm. RESULTS: Separable linear transformations allow for some data sets to decrease the error rate of the K-N classification rule based on the Euclidean distance. Such results can be seen on the example of data sets taken from the Hepar system of diagnosis support. CONCLUSIONS: Medical diagnosis support based on the CBR or the K-NN rules can be improved through separable transformations of the referencing sets.
Asunto(s)
Diagnóstico , Modelos Estadísticos , Algoritmos , Humanos , PoloniaRESUMEN
The "Hepar" system comprises a clinical database and the shell of procedures that aim at data analysis and the support of diagnosis. The database consists of hepatological patient cases. Each case is described by about 200 medical findings and histopathologically verified diagnosis. The diagnosis supporting rules of "Hepar" are based on visualizing data transformations and on the similarity based techniques. The applied linear visualizing transformations of data sets on the plane aim at separating of the groups of patients associated with different diseases. The resulting diagnostic maps by the visual inspection allow to find such cases in the database that are similar to the previously diagnosed patients. This paper examines combining of data transformations with the nearest neighbors techniques in the support of diagnosis. We report the results on the experimental comparisons of different decision rules including the feature selection procedure.
Asunto(s)
Técnicas de Apoyo para la Decisión , Diagnóstico por Computador , Sistemas Especialistas , Humanos , Hepatopatías/diagnóstico , Matemática , Redes Neurales de la ComputaciónRESUMEN
A new method of solving the problem of medical tests selection is proposed. This method is based on the notion of linear nonseparability of data. A measure of linear nonseparability is computed by minimization of the perceptron criterion function. A multistage strategy of search of the minimum sufficient tests subset is described. This strategy is reliable in the sense that it allows us to find the least (globally) sufficient number of tests. The considerations are illustrated on a liver diseases data example.
Asunto(s)
Hepatopatías/diagnóstico , Diagnóstico Diferencial , Errores Diagnósticos , Estudios de Evaluación como Asunto , Humanos , Pruebas de Función Hepática , Estadística como AsuntoRESUMEN
The binary decision element described by the decision rule depending upon weight vector w is a model of neuron examined in this paper. The environment of the element is described by some unknown, stationary distribution (p(kappa). The input signals kappa[n] of the element appear in each step n independently in accordance with the distribution p(kappa). During an unsupervised learning process the weight vector w[n] is changed on the base of the input vector kappa[n]. In the paper there are regarded two self-learning algorithms which are stochastic approximation type. For both algorithms the same rule of past experiences neglecting or the rule of weight decrease has been introduced. The first algorithm differs from the other one by a rule of weight increase. It has been proved that only one of these algorithms always leads to the same decision rule in a given environment p(kappa).
Asunto(s)
Aprendizaje , Modelos Neurológicos , Toma de Decisiones , Humanos , ProbabilidadRESUMEN
An algorithm of learning in multilayer threshold nets without feedbacks is proposed. The net is built of threshold elements with binary inputs. During a learning process each input vector chi is accompanied by a teacher's decision omega (omega epsilon(1,...,M)). The pairs (chi[n], omega[n]) appear in successive steps independently according to some unknown stationary distribution p(chi, omega). The problem of learning of a threshold net has been decomposed to a series of problems of learning of the threshold elements. The proposed learning algorithm of the threshold elements has a perceptron-like form. It was proven that a decision rule of the threshold net stabilizes after a finite number of steps. For definite classes (p(chi,omega))K of distributions p(chi, omega), an optimal decision rule stabilizes after a finite number of steps. These classes (p(chi, omega))K also contain distributions describing learning processes with perturbations.