Showing 8 results for Neural Network
A Biglarian, E Hajizadeh, A Kazemnejad,
Volume 6, Issue 3 (12-2010)
Abstract
Background & Objective: Using parametric models is common approach in survival analysis. In the recent years, artificial neural network (ANN) models have increasingly used in survival prediction. The aim of this study was to predict of survival rate of patients with gastric cancer by using a parametric regression and ANN models and compare these methods.
Methods: We used the data of 436 gastric cancer patients from a cancer registry in Tehran between 2002-2007. All patients had a confirmed diagnosis. Data were randomly divided into two groups: training and testing (or validation) set. For analysis of data we used a parametric model (exponential, Weibull, normal, lognormal, logistic and log-logistic models) and a three layer ANN model. In order to compare of the prediction of two models, we used the area under receiver operating characteristic (AUROC) curve, classification table and concordance index.
Results: The prediction accuracy of the ANN and the parametric (Weibull) models were 79.45% and 73.97% respectively. The AUROC for the ANN and the Weibull models were 0.815 and 0.748 respectively.
Conclusions: The ANN had a better predictions than the Weibull model. Thus it is suggested to use of the ANN model survival prediction in field of cancer.
M Sedehi, Y Mehrabi, A Kazemnejad, V Joharimajd, F Hadaegh,
Volume 6, Issue 4 (3-2011)
Abstract
Background & Objective: Mixed outcomes arise when, in a multivariate model, response variables measured on different scales such as binary and continuous. Artificial neural networks (ANN) can be used for modeling in situations where classic models have restricted application when some of their assumptions are not met. In this paper, we propose a method based on ANNs for modeling mixed binary and continuous outcomes.
Methods: Univariate and bivariate models were evaluated based on two different sets of simulated data. The scaled conjugate gradient (SCG) algorithm was used for optimization. To end the algorithm and finding optimum number of iteration and learning coefficient, mean squared error (MSE) was computed. Predictive accuracy rate criterion was employed for selection of appropriate model. We also used our model in medical data for joint prediction of metabolic syndrome (binary) and HOMA-IR (continues) in Tehran Lipid and Glucose Study (TLGS). The codes were written in R 2.9.0 and MATLAB 7.6.
Results: The predictive accuracy for univariate and bivariate models based on simulated dataset Ι, where two outcomes associated with a common covariate, were shown to be approximately similar. However, in simulated dataset ΙΙ in which two outcomes associated with different covariates, predictive accuracy in bivariate models were seen to be larger than that of univariate models.
Conclusions: It is indicated that the predictive accuracy gain is higher in bivariate model, when the outcomes share a different set of covariates with higher level of correlation between the outcomes.
A Falahati , K Soheili , M Nazifi , S Abbaspour,
Volume 9, Issue 2 (10-2013)
Abstract
Background & Objectives: Economic growth has been along with increasing energy demand in the world in addition environment pollutions which healthy life nowadays faces up with major challenges. Since there are several influential factors in this model, therefore this study designed to assess the effect of some independent socio-economic variables on the people health.
Methods: An artificial neural network (ANN) was developed to review health risk factors during the years 1971-2009. Using neural network methods in the study or the MPL method is .multi-layer perception.
Results: In ANN selected for this study, one hidden layer with three nodes is selected. Being more important the urban variable in modeling shows that the positive effect of urbanization on the health is more powerful than negative effects of air pollution.
Conclusion: Based on our model it is concluded that urbanization as a major risk factor to produce (accelerate) of air pollution, has the most negative effect on health and life expectancy.
A Asadabadi , A Bahrampour, Aa Haghdoost,
Volume 10, Issue 3 (12-2014)
Abstract
Background and Objectives : recent years, considerable attention has been paid to statistical models for classification of medical data according to various diseases and their outcomes. Artificial neural networks have been successfully used for pattern recognition and prediction since they are not based on prior assumptions in clinical studies. This study compared two statistical models, artificial neural network and logistic regression, to predict the survival of patients with breast cancer.
Methods: Two models were applied on cancer registry data, Kerman, southeast of Iran, to predict survival. The data of 712 breast cancer patients in the age group 15 to 85 years was used in this study. The logistic regression and three-layer perceptron neural network models were compared in terms of predicting the survival. Sensitivity, specificity, prediction accuracy, and the area under ROC curve were used for comparing the two models.
Results : In this study, the sensitivity and specificity of logistic regression and artificial neural network models were (0.594, 0.70) and (0.621, 0.723), respectively. Prediction accuracy and the area under ROC curve for two models were (0.688, 0.725) and (0.70, 0.725), respectively.
Conclusion: Although there were insignificant differences in the performance of the two models for predicting the survival of the patients with breast cancer, the corresponding results of artificial neural network were more appropriate for predicting survival in such data.
M Jahani, J Rezaenoor, E Hadavandi, I Salehi, H Tahsini,
Volume 11, Issue 2 (9-2015)
Abstract
Background & Objectives: In recent years, different decision support systems (DSS) have been used to predict and diagnose diseases. The purpose of this paper was to compare some DSSs and to evaluate their accuracy in predicting diabetes.
Methods: In this research, determination and optimization of the weights of the neural network were undertaken using genetic algorithm and Levenberg-Marquardt (GALM). Traditional and K-Fold Cross Validation were used to verify the models. Finally, the proposed model (i.e. GALM) was compared using logistic regression and genetic algorithm based on area under curve (AUC), and Confusion Matrix.
Results: After evaluating the results, the model based on the GALM algorithm showed better sensitivity and specificity in comparison with models based on the logistic regression (LR) and genetic algorithm (GA). Furthermore, among other models, the proposed model had a high sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and a small negative likelihood.
Conclusion: The results showed that the GALM model with a sensitivity, specificity, PPV, NPV, and AUC of 98.7, 90.01, 91.8, 98.3 and 0.979 respectively was an appropriate model for predicting diabetes in comparison with models of GA and LR.
L Tapak, N Shirmohammadi-Khorram , O Hamidi, Z Maryanaji,
Volume 14, Issue 2 (9-2018)
Abstract
Background and Objectives: Identification of statistical models has a great impact on early and accurate detection of outbreaks of infectious diseases and timely warning in health surveillance. This study evaluated and compared the performance of the three data mining techniques in time series prediction of brucellosis.
Methods: In this time series, the data of the human brucellosis cases and climatology parameters of Hamadan, west of Iran, were analyzed on a monthly basis from 2004 (March/April) to 2017 (February/March). The data were split into two subsets of train (80%) and test (20%). Three techniques, i.e. radial basis function (RBF) and multilayer perceptron (MLP) artificial neural network methods as well as K Nearest neighbor (KNN), were used in both subsets. The root mean square errors (RMSE), mean absolute errors (MAE), mean absolute relative errors (MARE), determination coefficient (R2) and intra-class correlation coefficient (ICC) were used for performance comparison.
Results: Results indicated that RMSE (23.79), MAE (20.65) and MARE (0.25) for MLP were smaller compared to the values of the other two models. The ICC (0.75) and R2 (0.61) values were also better for this model. Thus, the MLP model outperformed the other models in predicting the used data. The most important climatology variable was temperature.
Conclusion: MLP can be effectively applied to diagnose the behavior of brucellosis over time. Further research is necessary to detect the most suitable method for predicting the trend of this disease.
M Javanbakht, M Argani, K Ezimand, A Saghafipour,
Volume 17, Issue 1 (5-2021)
Abstract
Background and Objectives: Environmental conditions in different geographical areas provide a basis for the spread of some diseases. Cutaneous leishmaniasis is a serious threat to public health and is one of the arthropod-borne diseases. The prevalence and distribution of this disease is affected by environmental and climatic factors. The aim of this study was to model the Spatio-temporal variations in the incidence rate of this disease based on environmental and ecological criteria.
Methods: The northeast of Iran was selected as the study area. The data used in this study included vegetation, surface temperature, precipitation, evapotranspiration, soil moisture, digital elevation model and sunny hours. The artificial neural network method was used to model the spatio-temporal changes of cutaneous leishmaniasis.
Results: Spatial variations in the incidence of the disease had a north-south trend and decreased from north to south. In addition, two foci were identified in the medium altitude areas in North and South Khorasan provinces. Temporal variations in the incidence of disease in the study period showed that the incidence rate decreased in the two identified foci from 2011 to 2016.
Conclusion: The modeling results showed that the estimated regression coefficient was 0.92 for neural network based on all three types of data (training, validation, test) indicating good quality of constructed neural network. In addition, sensitivity analysis results showed that sunny hours and soil moisture were the most important factors in the model function.
Nasrin Talkhi, Nooshin Akbari Sharak, Zahra Rajabzadeh, Maryam Salari, Seyed Masoud Sadati, Mohammad Taghi Shakeri,
Volume 18, Issue 3 (12-2022)
Abstract
Background and Objectives: Due to the high prevalence of COVID-19 disease and its high mortality rate, it is necessary to identify the symptoms, demographic information and underlying diseases that effectively predict COVID-19 death. Therefore, in this study, we aimed to predict the mortality behavior due to COVID-19 in Khorasan Razavi province.
Methods: This study collected data from 51, 460 patients admitted to the hospitals of Khorasan Razavi province from 25 March 2017 to 12 September 2014. Logistic regression and Neural network methods, including machine learning methods, were used to identify survivors and non-survivors caused by COVID-19.
Results: Decreased consciousness, cough, PO2 level less than 93%, age, cancer, chronic kidney diseases, fever, headache, smoking status, and chronic blood diseases are the most important predictors of death. The accuracy of the artificial neural network model was 89.90% in the test phase. Also, the sensitivity, specificity and area under the rock curve in this model are equal to 76.14%, 91.99% and 77.65%, respectively.
Conclusion: Our findings highlight the importance of some demographic information, underlying diseases, and clinical signs in predicting survivors and non-survivors of COVID-19. Also, the neural network model provided high accuracy in prediction. However, medical research in this field will lead to complementary results by using other methods of machine learning and their high power.
