Implementation of Artificial Neural Networks and Decision Tree Algorithms for Heart Disease Diagnosis

Background: Various indices derived from red blood cell (RBC) parameters have been described for distinguishing betathalassemia minor and other types of hypochromic microcytic anemia. Objective: The study is aimed at investigating thediagnostic reliability of different RBC indices and formulas in differentiation between beta thalassemia minor and othertypes of hypochromic microcytic anemia. Subjects and Methods: This is a cross‐sectional study which was carried out sincefirst of Jan 2011 to end of December 2011 on 171 children with hypochromic microcytic anemia in Kut Oncology Centre,Wasit, Iraq. Results: There was a statistical significant difference between thalassemic group and other groups regardingblood indices as well as the eight formulas which were used. The highest correctly identified patients (PCIP) was reportedfor RBCs count (84%) with sensitivity and specificity of 96.3%. The Youden's index for RBCs was 58.2 which is the highestvalue compared with other seven parameters or indices which were used in this study. The second highest Youden's indexwas for G & K index, with 78.4% PCIP, and sensitivity and specificity of 98.2%. Youden's index of red cell distributionwidth (RDW) was the lowest value compared to other values used in this study as well as the lowest percentage of correctlyidentified patients (65%). The sensitivity and specificity of RDW for BTM was 86.1%. Conclusion: According to this study,cell counter-based parameters and formulas, particularly RBCs, and Green and King index are superior to all othermethods examined for distinguishing between thalassemia trait and other hypochromic microcytic anemia; while, RDW wasinadequate and ineffective for that purpose.

Classification Algorithm Performance Study on Diabetes Electronic Medical Records / 医学信息学杂志

The paper preprocesses the data including basic information,admission and discharge record and progress note of diabetes Electronic Medical Records (EMR),implementing decision tree,Artificial Neural Network (ANN),Naive bayesian and K-Nearest Neighbor (KNN) classifications respectively on data that have been processed with Weka 3.9.The result shows that Naive bayesian classification,which is superior to the others in predicting and classifying such data,can provide basis for the classification and prediction of diabetes.

Application of Association Rules in Data Mining of Circulation Library Readers / 医学信息学杂志

〔Abstract〕 The paper introduces the application of association rules in the library .With circulation data of freshmen in different ma-jors, in terms of different majors and grades of the college , it extracts partial records suitable for data mining , analyzes phases of data preparation and data mining by use of association rules , and provides technical support for in -depth conduction of reader services .

Mining the relation between leukemia and genes using Weka / 中华医学图书情报杂志

Objective To mine the relation between leukemia and genes using Weka. Methods The papers on leuke-mia and genes were retrieved from PubMed, their subject headings and subheadings were extracted using BICOMB to generate co-occurrence matrix and term-paper matrix. The research hotspots were found by cluster analysis of the data on co-occurrence matrix using Weka and Cobweb. The literature was verified. Results The 42 high fre-quency words were clustered into 7 classes by Weka. No high frequency words of leukemia or genes were found in classes 1, 2, 4 and 5, indicating that their clustering efficiency was poor. The clustering efficiency of the other 3 classes was good. Conclusion Cluster analysis showed that leukemia is related with myc gene, ab1 gene, p53 gene, virus gene, immunoglobulin gene and mdm gene.

Técnicas de minería de datos aplicadas al diagnóstico de entidades clínicas / Data mining techniques aplied to diagnosys of clinical entities

Disminuir el error médico y mejorar los procesos de salud es prioridad de todo el personal sanitario. En este contexto surgen los Sistemas Clínicos de Soporte para la Toma de Decisiones (CDSS), los cuales son un componente fundamental en la informatización de la capa clínica. Con la evolución de las tecnologías gran cantidad de datos han podido ser estudiados y clasificados a partir de la minería de datos. Una de las principales ventajas de la utilización de esta, en los CDSS, ha sido su capacidad de generar nuevos conocimientos. Con este fin se propone, mediante la combinación de dos modelos matemáticos, cómo se puede contribuir al diagnóstico de enfermedades usando técnicas de minería de datos. Para mostrar los modelos utilizados se tomó como caso de estudio la hipertensión arterial. El desarrollo de la investigación se rige por la metodología más utilizada actualmente en los procesos de Descubrimiento de Conocimiento en Bases de Datos: CRISP-DM 1.0, y se apoya en la herramienta de libre distribución WEKA 3.6.2, de gran prestigio entre las utilizadas para el modelado de minería de datos. Como resultados se obtuvieron diversos patrones de comportamiento con relación a los factores de riesgo a sufrir hipertensión mediante técnicas de minería de datos(AU)

Reduce medical errors and improve health processes is a priority of all health personnel. In this context arise the Clinical Support Systems for Decision Making (CDSS), which are a key component in computerization of the clinical layer. With the evolution of technologies, large amounts of data have been studied and classified based on data mining. One of the main advantages of using this in the CDSS, has been its ability to generate new knowledge. For this purpose, this paper presents, by combining two mathematical models, a way to contribute to the diagnosis of diseases using data mining techniques. Hypertension was taken as a case study to show the models used. The research development methodology follows the most used processes of knowledge discovery in databases: CRISP-DM 1.0, and relies on the free distribution tool WEKA 3.6.2. We obtained different patterns of behavior in relation to risk factors for developing hypertension using data mining techniques(AU)