Displaying publications 21 - 40 of 421 in total

Abstract:
Sort:
  1. Modelling exchange rates using regime switching models
    Mohd Tahir Ismail, Zaidi Isa
    Sains Malaysiana, 2006;35:55-62.
    The behaviour of many financial time series cannot be modeled solely by linear time series model. Phenomena such as mean reversion, volatility of stock markets and structural breaks cannot be modelled implicitly using simple linear time series model. Thus, to overcome this problem, nonlinear time series models are typically designed to accommodate these nonlinear features in the data. In this paper, we use portmanteau test and structural change test to detect nonlinear feature in three ASEAN countries exchange rates (Malaysia, Singapore and Thailand). It is found that the null hypothesis of linearity is rejected and there is evidence of structural breaks in the exchange rates series. Therefore, the decision of using regime switching model in this study is justified. Using model selection criteria (AIC, SBC, HQC), we compare the in-sample fitting between two types of regime switching model. The two regime switching models we considered were the Self-Exciting Threshold Autoregressive (SETAR) model and the Markov switching Autoregressive (MS-AR) model where these models can explain the abrupt changes in a time series but differ as how they model the movement between regimes. From the AIC, SBC and HQC values, it is found that the MS -AR model is the best fitted model for all the return series. In addition, the regime switching model also found to perform better than simple autoregressive model in in-sample fitting. This result justified that nonlinear model give better in-sample fitting than linear model.
  2. Neuro-genetic optimization of temperature control for a continuous flow polymerase chain reaction microdevice
    Lee HW, Arunasalam P, Laratta WP, Seetharamu KN, Azid IA
    J Biomech Eng, 2007 Aug;129(4):540-7.
    PMID: 17655475
    In this study, a hybridized neuro-genetic optimization methodology realized by embedding finite element analysis (FEA) trained artificial neural networks (ANN) into genetic algorithms (GA), is used to optimize temperature control in a ceramic based continuous flow polymerase chain reaction (CPCR) device. The CPCR device requires three thermally isolated reaction zones of 94 degrees C, 65 degrees C, and 72 degrees C for the denaturing, annealing, and extension processes, respectively, to complete a cycle of polymerase chain reaction. The most important aspect of temperature control in the CPCR is to maintain temperature distribution at each reaction zone with a precision of +/-1 degree C or better, irrespective of changing ambient conditions. Results obtained from the FEA simulation shows good comparison with published experimental work for the temperature control in each reaction zone of the microfluidic channels. The simulation data are then used to train the ANN to predict the temperature distribution of the microfluidic channel for various heater input power and fluid flow rate. Once trained, the ANN analysis is able to predict the temperature distribution in the microchannel in less than 20 min, whereas the FEA simulation takes approximately 7 h to do so. The final optimization of temperature control in the CPCR device is achieved by embedding the trained ANN results as a fitness function into GA. Finally, the GA optimized results are used to build a new FEA model for numerical simulation analysis. The simulation results for the neuro-genetic optimized CPCR model and the initial CPCR model are then compared. The neuro-genetic optimized model shows a significant improvement from the initial model, establishing the optimization method's superiority.
    Matched MeSH terms: Neural Networks (Computer)*
  3. Automated detection of anesthetic depth levels using chaotic features with artificial neural networks
    Lalitha V, Eswaran C
    J Med Syst, 2007 Dec;31(6):445-52.
    PMID: 18041276
    Monitoring the depth of anesthesia (DOA) during surgery is very important in order to avoid patients' interoperative awareness. Since the traditional methods of assessing DOA which involve monitoring the heart rate, pupil size, sweating etc, may vary from patient to patient depending on the type of surgery and the type of drug administered, modern methods based on electroencephalogram (EEG) are preferred. EEG being a nonlinear signal, it is appropriate to use nonlinear chaotic parameters to identify the anesthetic depth levels. This paper discusses an automated detection method of anesthetic depth levels based on EEG recordings using non-linear chaotic features and neural network classifiers. Three nonlinear parameters, namely, correlation dimension (CD), Lyapunov exponent (LE) and Hurst exponent (HE) are used as features and two neural network models, namely, multi-layer perceptron network (feed forward model) and Elman network (feedback model) are used for classification. The neural network models are trained and tested with single and multiple features derived from chaotic parameters and the performances are evaluated in terms of sensitivity, specificity and overall accuracy. It is found from the experimental results that the Lyapunov exponent feature with Elman network yields an overall accuracy of 99% in detecting the anesthetic depth levels.
    Matched MeSH terms: Neural Networks (Computer)*
  4. Comparison of estimation capabilities of response surface methodology (RSM) with artificial neural network (ANN) in lipase-catalyzed synthesis of palm-based wax ester
    Basri M, Rahman RN, Ebrahimpour A, Salleh AB, Gunawan ER, Rahman MB
    BMC Biotechnol, 2007;7:53.
    PMID: 17760990
    Wax esters are important ingredients in cosmetics, pharmaceuticals, lubricants and other chemical industries due to their excellent wetting property. Since the naturally occurring wax esters are expensive and scarce, these esters can be produced by enzymatic alcoholysis of vegetable oils. In an enzymatic reaction, study on modeling and optimization of the reaction system to increase the efficiency of the process is very important. The classical method of optimization involves varying one parameter at a time that ignores the combined interactions between physicochemical parameters. RSM is one of the most popular techniques used for optimization of chemical and biochemical processes and ANNs are powerful and flexible tools that are well suited to modeling biochemical processes.
    Matched MeSH terms: Neural Networks (Computer)*
  5. Extending the range of an optical vanadium(v) sensor based on immobilized fatty hydroxamic acid in poly (methyl methacrylate) using artificial neural network
    Azizul Isha, Nor Azah Yusof, Musa Ahmad, Dedy Suhendra, Wan Md. Zin Wan Yunus, Zulkarnain Zainal
    Pertanika Journal of Science & Technology, 2007;15(2):121-130.
    MyJurnal
    An artificial neural network (ANN) was applied for the determination of V(V) based on immobilized fatty hydroxamic acid (FHA) in poly(methyl methacrylate) (PMMA). Spectra obtained from the V(V)-FHA complex at single wavelengths was used as the input data for the ANN. The V(V)-FHA complex shows a limited linear dynamic range of V(V) concentration of 10 - 100 mg/ L. After training with ANN, the linear dynamic range was extended with low calibration error. A three layer feed forward neural network using backpropagation (BP) algorithm was employed in this study. The input layer consisted of single neurons, 30 neurons in hidden a layer and one output neuron was found appropriate for the multivariate calibration used. The network were trained up to 10000 epochs with 0.003 % learning rate. This reagent also provided a good analytical pedormance with reproducibility characters of the method yielding relative standard deviation (RSD) of 9.29% and 7.09% for V(V) at concentrations of 50 mg/ L and 200 mg/ L, respectively. The limit of detection of the method was 8.4 mg/ L.
  6. An automated cervical pre-cancerous diagnostic system
    Mat-Isa NA, Mashor MY, Othman NH
    Artif Intell Med, 2008 Jan;42(1):1-11.
    PMID: 17996432
    This paper proposes to develop an automated diagnostic system for cervical pre-cancerous. METHODS AND DATA SAMPLES: The proposed automated diagnostic system consists of two parts; an automatic feature extraction and an intelligent diagnostic. In the automatic feature extraction, the system automatically extracts four cervical cells features (i.e. nucleus size, nucleus grey level, cytoplasm size and cytoplasm grey level). A new features extraction algorithm called region-growing-based features extraction (RGBFE) is proposed to extract the cervical cells features. The extracted features will then be fed as input data to the intelligent diagnostic part. A new artificial neural network (ANN) architecture called hierarchical hybrid multilayered perceptron (H(2)MLP) network is proposed to predict the cervical pre-cancerous stage into three classes, namely normal, low grade intra-epithelial squamous lesion (LSIL) and high grade intra-epithelial squamous lesion (HSIL). We empirically assess the capability of the proposed diagnostic system using 550 reported cases (211 normal cases, 143 LSIL cases and 196 HSIL cases).
  7. A hybrid ART-GRNN online learning neural network with a epsilon -insensitive loss function
    Yap KS, Lim CP, Abidin IZ
    IEEE Trans Neural Netw, 2008 Sep;19(9):1641-6.
    PMID: 18779094 DOI: 10.1109/TNN.2008.2000992
    In this brief, a new neural network model called generalized adaptive resonance theory (GART) is introduced. GART is a hybrid model that comprises a modified Gaussian adaptive resonance theory (MGA) and the generalized regression neural network (GRNN). It is an enhanced version of the GRNN, which preserves the online learning properties of adaptive resonance theory (ART). A series of empirical studies to assess the effectiveness of GART in classification, regression, and time series prediction tasks is conducted. The results demonstrate that GART is able to produce good performances as compared with those of other methods, including the online sequential extreme learning machine (OSELM) and sequential learning radial basis function (RBF) neural network models.
    Matched MeSH terms: Neural Networks (Computer)*
  8. Fulltext A modeling study by response surface methodology and artificial neural network on culture parameters optimization for thermostable lipase production from a newly isolated thermophilic Geobacillus sp. strain ARM
    Ebrahimpour A, Abd Rahman RN, Ean Ch'ng DH, Basri M, Salleh AB
    BMC Biotechnol, 2008 Dec 23;8:96.
    PMID: 19105837 DOI: 10.1186/1472-6750-8-96
    BACKGROUND: Thermostable bacterial lipases occupy a place of prominence among biocatalysts owing to their novel, multifold applications and resistance to high temperature and other operational conditions. The capability of lipases to catalyze a variety of novel reactions in both aqueous and nonaqueous media presents a fascinating field for research, creating interest to isolate novel lipase producers and optimize lipase production. The most important stages in a biological process are modeling and optimization to improve a system and increase the efficiency of the process without increasing the cost.

    RESULTS: Different production media were tested for lipase production by a newly isolated thermophilic Geobacillus sp. strain ARM (DSM 21496 = NCIMB 41583). The maximum production was obtained in the presence of peptone and yeast extract as organic nitrogen sources, olive oil as carbon source and lipase production inducer, sodium and calcium as metal ions, and gum arabic as emulsifier and lipase production inducer. The best models for optimization of culture parameters were achieved by multilayer full feedforward incremental back propagation network and modified response surface model using backward elimination, where the optimum condition was: growth temperature (52.3 degrees C), medium volume (50 ml), inoculum size (1%), agitation rate (static condition), incubation period (24 h) and initial pH (5.8). The experimental lipase activity was 0.47 Uml(-1) at optimum condition (4.7-fold increase), which compared well to the maximum predicted values by ANN (0.47 Uml(-1)) and RSM (0.476 Uml(-1)), whereas R2 and AAD were determined as 0.989 and 0.059% for ANN, and 0.95 and 0.078% for RSM respectively.

    CONCLUSION: Lipase production is the result of a synergistic combination of effective parameters interactions. These parameters are in equilibrium and the change of one parameter can be compensated by changes of other parameters to give the same results. Though both RSM and ANN models provided good quality predictions in this study, yet the ANN showed a clear superiority over RSM for both data fitting and estimation capabilities. On the other hand, ANN has the disadvantage of requiring large amounts of training data in comparison with RSM. This problem was solved by using statistical experimental design, to reduce the number of experiments.

    Matched MeSH terms: Neural Networks (Computer)*
  9. Model-based neuroimaging for cognitive computing
    Poznanski RR
    J Integr Neurosci, 2009 Sep;8(3):345-69.
    PMID: 19938210
    The continuity of the mind is suggested to mean the continuous spatiotemporal dynamics arising from the electrochemical signature of the neocortex: (i) globally through volume transmission in the gray matter as fields of neural activity, and (ii) locally through extrasynaptic signaling between fine distal dendrites of cortical neurons. If the continuity of dynamical systems across spatiotemporal scales defines a stream of consciousness then intentional metarepresentations as templates of dynamic continuity allow qualia to be semantically mapped during neuroimaging of specific cognitive tasks. When interfaced with a computer, such model-based neuroimaging requiring new mathematics of the brain will begin to decipher higher cognitive operations not possible with existing brain-machine interfaces.
    Matched MeSH terms: Neural Networks (Computer)*
  10. Application of artificial neural network for yield prediction of lipase-catalyzed synthesis of dioctyl adipate
    Abdul Rahman MB, Chaibakhsh N, Basri M, Salleh AB, Abdul Rahman RN
    Appl Biochem Biotechnol, 2009 Sep;158(3):722-35.
    PMID: 19132557 DOI: 10.1007/s12010-008-8465-z
    In this study, an artificial neural network (ANN) trained by backpropagation algorithm, Levenberg-Marquadart, was applied to predict the yield of enzymatic synthesis of dioctyl adipate. Immobilized Candida antarctica lipase B was used as a biocatalyst for the reaction. Temperature, time, amount of enzyme, and substrate molar ratio were the four input variables. After evaluating various ANN configurations, the best network was composed of seven hidden nodes using a hyperbolic tangent sigmoid transfer function. The correlation coefficient (R2) and mean absolute error (MAE) values between the actual and predicted responses were determined as 0.9998 and 0.0966 for training set and 0.9241 and 1.9439 for validating dataset. A simulation test with a testing dataset showed that the MAE was low and R2 was close to 1. These results imply the good generalization of the developed model and its capability to predict the reaction yield. Comparison of the performance of radial basis network with the developed models showed that radial basis function was more accurate but its performance was poor when tested with unseen data. In further part of the study, the feedforward backpropagation model was used for prediction of the ester yield within the given range of the main parameters.
    Matched MeSH terms: Neural Networks (Computer)*
  11. An intelligent load shedding scheme using neural networks and neuro-fuzzy
    Haidar AM, Mohamed A, Al-Dabbagh M, Hussain A, Masoum M
    Int J Neural Syst, 2009 Dec;19(6):473-9.
    PMID: 20039470
    Load shedding is some of the essential requirement for maintaining security of modern power systems, particularly in competitive energy markets. This paper proposes an intelligent scheme for fast and accurate load shedding using neural networks for predicting the possible loss of load at the early stage and neuro-fuzzy for determining the amount of load shed in order to avoid a cascading outage. A large scale electrical power system has been considered to validate the performance of the proposed technique in determining the amount of load shed. The proposed techniques can provide tools for improving the reliability and continuity of power supply. This was confirmed by the results obtained in this research of which sample results are given in this paper.
    Matched MeSH terms: Neural Networks (Computer)*
  12. Drug-like and non drug-like pattern classification based on simple topology descriptor using hybrid neural network
    Wan-Mamat WM, Isa NA, Wahab HA, Wan-Mamat WM
    Annu Int Conf IEEE Eng Med Biol Soc, 2009;2009:6424-7.
    PMID: 19964424 DOI: 10.1109/IEMBS.2009.5333747
    An intelligent prediction system has been developed to discriminate drug-like and non drug-like molecules pattern. The system is constructed by using the application of advanced version of standard multilayer perceptron (MLP) neural network called Hybrid Multilayer Perceptron (HMLP) neural network and trained using Modified Recursive Prediction Error (MRPE) training algorithm. In this work, a well understood and easy excess Rule of Five + Veber filter properties are selected as the topological descriptor. The main idea behind the selection of this simple descriptor is to assure that the system could be used widely, beneficial and more advantageous regardless at all user level within a drug discovery organization.
    Matched MeSH terms: Neural Networks (Computer)*
  13. Fulltext Model development of microwave remote sensing of sea ice
    Lee, Y.J., Yap, H.J., Lim, W.K., Ewe, H.T., Chuah, H.T.
    ASM Science Journal, 2009;3(2):131-142.
    MyJurnal
    Three techniques to retrieve information on sea ice thickness from both active and passive radar backscatter data are presented. The first inversion model is a combination of the radiative transfer theory with dense medium phase and amplitude correction theory (DMPACT), and the Levenberg-Marquardt optimization algorithm. The radiative transfer theory was applied as the forward model to generate radar backscatter data, while the DMPACT was included to account for the close spacing effect among the scatterers within the medium. The Levenberg-Marquardt optimization algorithm was then applied to reduce the error between the model generated radar backscatter data and the measured radar backscatter data from satellite images so that the sea ice thickness could be estimated. The second method presented was the neural network inversion method which utilizes a chain of neurons with variable weights. Once the network was fully operational it would be possible to predict the sea ice thickness, provided sufficient training data are given. The last method was the genetic algorithm which is a search technique used in order to predict the approximate sea ice thickness from the measured data. Data from ground truth measurements carried out in Ross Island, Antarctica, together with radar backscatter data extracted from purchased satellite images were used as input to verify the models. All three models were tested and successfully predicted sea ice thickness from actual terrain using the ground truth measurement data, with several constraints and assumptions placed to avoid problems during the retrieval process. While the models still have their own limitations, the potential use of the models for actual sea ice thickness retrieval was confirmed.
  14. Spatial assessment of Langat River water quality using chemometrics
    Juahir H, Zain SM, Aris AZ, Yusoff MK, Mokhtar MB
    J Environ Monit, 2010 Jan;12(1):287-95.
    PMID: 20082024 DOI: 10.1039/b907306j
    The present study deals with the assessment of Langat River water quality with some chemometrics approaches such as cluster and discriminant analysis coupled with an artificial neural network (ANN). The data used in this study were collected from seven monitoring stations under the river water quality monitoring program by the Department of Environment (DOE) from 1995 to 2002. Twenty three physico-chemical parameters were involved in this analysis. Cluster analysis successfully clustered the Langat River into three major clusters, namely high, moderate and less pollution regions. Discriminant analysis identified seven of the most significant parameters which contribute to the high variation of Langat River water quality, namely dissolved oxygen, biological oxygen demand, pH, ammoniacal nitrogen, chlorine, E. coli, and coliform. Discriminant analysis also plays an important role as an input selection parameter for an ANN of spatial prediction (pollution regions). The ANN showed better prediction performance in discriminating the regional area with an excellent percentage of correct classification compared to discriminant analysis. Multivariate analysis, coupled with ANN, is proposed, which could help in decision making and problem solving in the local environment.
  15. Reexamination of risk criteria in dengue patients using the self-organizing map
    Faisal T, Taib MN, Ibrahim F
    Med Biol Eng Comput, 2010 Mar;48(3):293-301.
    PMID: 20016950 DOI: 10.1007/s11517-009-0561-x
    Even though the World Health Organization criteria's for classifying the dengue infection have been used for long time, recent studies declare that several difficulties have been faced by the clinicians to apply these criteria. Accordingly, many studies have proposed modified criteria to identify the risk in dengue patients based on statistical analysis techniques. None of these studies utilized the powerfulness of the self-organized map (SOM) in visualizing, understanding, and exploring the complexity in multivariable data. Therefore, this study utilized the clustering of the SOM technique to identify the risk criteria in 195 dengue patients. The new risk criteria were defined as: platelet count less than or equal 40,000 cells per mm(3), hematocrit concentration great than or equal 25% and aspartate aminotransferase (AST) rose by fivefold the normal upper limit for AST/alanine aminotransfansferase (ALT) rose by fivefold the normal upper limit for ALT. The clusters analysis indicated that any dengue patient fulfills any two of the risk criteria is consider as high risk dengue patient.
    Matched MeSH terms: Neural Networks (Computer)*
  16. Nonlinear process modeling of fructosyltransferase (FTase) using bootstrap re-sampling neural network model
    Ahmad Z, Don MM, Mortan SH, Noor RA
    Bioprocess Biosyst Eng, 2010 Jun;33(5):599-606.
    PMID: 19915872 DOI: 10.1007/s00449-009-0381-2
    Recently, the increased demand of fructooligosaccharides (FOS) as a functional food has alarmed researchers to screen and identify new strains capable of producing fructosyltransferase (FTase). FTase is the enzyme that converts the substrate (sucrose) to glucose and fructose. The characterization of complex sugar such as table sugar, brown sugar, molasses, etc. will be carried out and the sugar that contained the highest sucrose concentration will be selected as a substrate. Eight species of macro-fungi will be screened for its ability to produce FTase and only one strain with the highest FTase activity will be selected for further studies. In this work, neural networks (NN) have been chosen to model the process based on their excellent 'resume' in coping with nonlinear process. Bootstrap re-sampling method has been utilized in re-sampling the data in this work. This method has successfully modeled the process as shown in the results.
    Matched MeSH terms: Neural Networks (Computer)*
  17. Parameters optimization of rice husk ash (RHA)/CaO/CeO2 sorbent for predicting SO2/NO sorption capacity using response surface and neural network models
    Dahlan I, Ahmad Z, Fadly M, Lee KT, Kamaruddin AH, Mohamed AR
    J Hazard Mater, 2010 Jun 15;178(1-3):249-57.
    PMID: 20137857 DOI: 10.1016/j.jhazmat.2010.01.070
    In this work, the application of response surface and neural network models in predicting and optimizing the preparation variables of RHA/CaO/CeO(2) sorbent towards SO(2)/NO sorption capacity was investigated. The sorbents were prepared according to central composite design (CCD) with four independent variables (i.e. hydration period, RHA/CaO ratio, CeO(2) loading and the use of RHA(raw) or pretreated RHA(600 degrees C) as the starting material). Among all the variables studied, the amount of CeO(2) loading had the largest effect. The response surface models developed from CCD was effective in providing a highly accurate prediction for SO(2) and NO sorption capacities within the range of the sorbent preparation variables studied. The prediction of CCD experiment was verified by neural network models which gave almost similar results to those determined by response surface models. The response surface models together with neural network models were then successfully used to locate and validate the optimum hydration process variables for maximizing the SO(2)/NO sorption capacities. Through this optimization process, it was found that maximum SO(2) and NO sorption capacities of 44.34 and 3.51 mg/g, respectively could be obtained by using RHA/CaO/CeO(2) sorbents prepared from RHA(raw) with hydration period of 12h, RHA/CaO ratio of 2.33 and CeO(2) loading of 8.95%.
  18. Pore size prediction of polyethersulfone ultrafiltration membranes using artificial neural networks
    Ibrahim MZ, Norashikin MZ
    J Nanosci Nanotechnol, 2010 Sep;10(9):6211-5.
    PMID: 21133176
    This paper reports the performance of two different artificial neural networks (ANN), Multi Layer Perceptron (MLP) and Radial Basis Function (RBF) compared to conventional software for prediction of the pore size of the asymmetric polyethersulfone (PES) ultrafiltration membranes. ANN has advantages such as incredible approximation, generalization and good learning ability. The MLP are well suited for multiple inputs and multiple outputs while RBF are powerful techniques for interpolation in multidimensional space. Three experimental data sets were used to train the ANN using polyethylene glycol (PEG) of different molecular weights as additives namely as PEG 200, PEG 400 and PEG 600. The values of the pore size can be determined manually from the graph and solve it using mathematical equation. However, the mathematical solution used to determine the pore size and pore size distribution involve complicated equations and tedious. Thus, in this study, MLP and RBF are applied as an alternative method to estimate the pore size of polyethersulfone (PES) ultrafiltration membranes. The raw data needed for the training are solute separation and solute diameter. Values of solute separation were obtained from the ultrafiltration experiments and solute diameters ware calculated using mathematical equation. With the development of this ANN model, the process to estimate membrane pore size could be made easier and faster compared to mathematical solutions.
  19. Fault diagnosis of the polypropylene production process (UNIPOL PP) using ANFIS
    Lau CK, Heng YS, Hussain MA, Mohamad Nor MI
    ISA Trans, 2010 Oct;49(4):559-66.
    PMID: 20667537 DOI: 10.1016/j.isatra.2010.06.007
    The performance of a chemical process plant can gradually degrade due to deterioration of the process equipment and unpermitted deviation of the characteristic variables of the system. Hence, advanced supervision is required for early detection, isolation and correction of abnormal conditions. This work presents the use of an adaptive neuro-fuzzy inference system (ANFIS) for online fault diagnosis of a gas-phase polypropylene production process with emphasis on fast and accurate diagnosis, multiple fault identification and adaptability. The most influential inputs are selected from the raw measured data sets and fed to multiple ANFIS classifiers to identify faults occurring in the process, eliminating the requirement of a detailed process model. Simulation results illustrated that the proposed method effectively diagnosed different fault types and severities, and that it has a better performance compared to a conventional multivariate statistical approach based on principal component analysis (PCA). The proposed method is shown to be simple to apply, robust to measurement noise and able to rapidly discriminate between multiple faults occurring simultaneously. This method is applicable for plant-wide monitoring and can serve as an early warning system to identify process upsets that could threaten the process operation ahead of time.
  20. Non-invasive diagnosis of risk in dengue patients using bioelectrical impedance analysis and artificial neural network
    Ibrahim F, Faisal T, Salim MI, Taib MN
    Med Biol Eng Comput, 2010 Nov;48(11):1141-8.
    PMID: 20683676 DOI: 10.1007/s11517-010-0669-z
    This paper presents a new approach to diagnose and classify early risk in dengue patients using bioelectrical impedance analysis (BIA) and artificial neural network (ANN). A total of 223 healthy subjects and 207 hospitalized dengue patients were prospectively studied. The dengue risk severity criteria was determined and grouped based on three blood investigations, namely, platelet (PLT) count (less than or equal to 30,000 cells per mm(3)), hematocrit (HCT) (increase by more than or equal to 20%), and either aspartate aminotransferase (AST) level (raised by fivefold the normal upper limit) or alanine aminotransferase (ALT) level (raised by fivefold the normal upper limit). The dengue patients were classified according to their risk groups and the corresponding BIA parameters were subsequently obtained and quantified. Four parameters were used for training and testing the ANN which are day of fever, reactance, gender, and risk group's quantification. Day of fever was defined as the day of fever subsided, i.e., when the body temperature fell below 37.5°C. The blood investigation and the BIA data were taken for 5 days. The ANN was trained via the steepest descent back propagation with momentum algorithm using the log-sigmoid transfer function while the sum-squared error was used as the network's performance indicator. The best ANN architecture of 3-6-1 (3 inputs, 6 neurons in the hidden layer, and 1 output), learning rate of 0.1, momentum constant of 0.2, and iteration rate of 20,000 was pruned using a weight-eliminating method. Eliminating a weight of 0.05 enhances the dengue's prediction risk classification accuracy of 95.88% for high risk and 96.83% for low risk groups. As a result, the system is able to classify and diagnose the risk in the dengue patients with an overall prediction accuracy of 96.27%.
    Matched MeSH terms: Neural Networks (Computer)*
Related Terms
Filters
Contact Us

Please provide feedback to Administrator (afdal@afpm.org.my)

External Links