Employing machine learning models to predict pregnancy termination among adolescent and young women aged 15-24 years in East Africa

Pregnancy termination is still a sensitive and continuing public health issue due to several political, economic, religious, and social concerns. This study assesses the applications of machine learning models in the prediction of pregnancy termination using data from eleven national datasets in East Africa. Nine machine learning models, namely: Random Forests (RF), Decision Tree, Logistic Regression, Support Vector Machine, eXtreme Gradient Boosting (XGB), AdaBoost, CatBoost, K-nearest neighbor, and feedforward neural network models were used to predict pregnancy termination, with six evaluation criteria utilized to compare their performance. The pooled prevalence of pregnancy termination in East Africa was found to be 4.56%. All machine learning models had an accuracy of at least 71.8% on average. The RF model provided accuracy, specificity, precision, and AUC of 92.9%, 0.87, 0.91, and 0.93, respectively. The most important variables for predicting pregnancy termination were marital status, age, parity, country of residence, age at first sexual activity, exposure to mass media, and educational attainment. These findings underscore the need for a tailored approach that considers socioeconomic and regional disparities in designing policy initiatives aimed at reducing the rate of pregnancy terminations among younger women in the region.

An unsupervised approach for personalized RHM with reduced mean alert latency

Remote Health Monitoring (RHM) is an important research topic among the researchers, where many challenges are to be addressed with respect to communication, device, synchronization, data analysis, knowledge inferencing, database maintenance, security, timely notification etc. Among these multi challenges, personalization of health data and scheduling of alert generation have been focused on this work. Recognizing the regular health pattern of each individual helps in diagnosing the disease accurately (reduces the False Alarm Ratio (FAR)) and provides the necessary treatment earlier. Similarly, in real time, with multiple patients, the latency should be minimal for timely alert generation. To address these two challenges, a Density-based K- means clustering (DbK-meansC) approach has been proposed in this work that personalize the vital health values. From the personalized health values the abnormalities in the health status of a person can be detected earlier. Here the health records are continuously updated with respect to health values that reflects in personalization of health records. If any abnormality noted in the health values, then the proposed work sends an alert message to the caretaker / the respective doctor using a dynamic preemptive priority scheduling scheme. The scheduling is done with respect to the severity levels of the vital health values of each individual respectively. The arrived results show that the proposed personalized abnormality detection RHM model generate alerts with minimum latency in terms of response and waiting time in a multi patient environment. With proper personalization, the obtained specificity and sensitivity are 91.56% and 92.87% respectively and the computational time is reduced as the degree of personalization increases.

Recurrent Neural Network-Based Multimodal Deep Learning for Estimating Missing Values in Healthcare

This estimation method operates by integrating the input values that are redundantly collected from heterogeneous devices through the selection of a representative value and estimating missing values by using a multimodal RNN. Users use a heterogeneous healthcare platform mainly in a mobile environment. Users who pay a relatively large amount of attention to healthcare possess various types of healthcare devices and collect data through their mobile devices. The collected data may be duplicated depending on the types of these devices. This data duplication causes an ambiguity issue in that it is difficult to determine which value among multiple data should be taken as the user’s actual value. Accordingly, it is necessary to create a neural network structure that considers the data value at the time previous to the current time. RNNs are appropriate for handling data with a time series characteristic. To learn an RNN-based neural network, learning data that have the same time step are required. Therefore, an RNN in which one variable becomes single-modal was designed for each learning run. In the RNN, a cell is a gated recurrent unit (GRU) cell that presents sufficient accuracy in the small resource environment of mobile devices. The RNNs that are learned according to the variables can each operate without additional learning, even if the situation of the user’s mobile device changes. In a heterogeneous environment, missing values are generated by various types of errors, including errors caused by battery charge and discharge, sensor failure, equipment exchange, and near-field communication errors. The higher the missing value ratio, the greater the number of errors that are likely to occur. For this reason, to achieve a more stable heterogeneous health platform, missing values must be considered. In this study, a missing value was estimated by means of multimodal deep learning; that is, a multimodal deep learning method was designed with one neural network that was connected with each learned single-modal RNN using a fully connected network (FCN). Each RNN input value delivers mutual influence through the weights of the FCN, and thereby, it is possible to estimate an output value even if any one of the input values is missing. According to the evaluation in terms of representative value selection, when a representative value was selected by using the mean or median, the most stable service was achieved. As a result of the evaluation according to the estimation method, the accuracy of the RNN-based multimodal deep learning method is 3.91%p higher than that of the SVD method.

Customer Relationship Management Based on SPRINT Classification Algorithm under Data Mining Technology

Under the advance of computational intelligence, customer relationship management system based on data mining technology can not only bring more economic benefits to an enterprise but also improve the management and decision-making level of Chinese enterprises. In this paper, the application of data mining technology in customer relationship management (CRM) is analyzed, and four data mining modes are realized: customer classification, cross-marketing, customer acquisition, and customer retention. In the data mining module, SPRINT classification algorithm is used in customer classification. At the same time, FP-growth, an association rule algorithm without candidate set, is applied in cross-marketing, which enhances the practicability of the system. The algorithm of optimal customer retention strategy under digital intelligence technology is adopted in customer retention, which makes up for the shortcomings of traditional CRM system and helps enterprises to better operate and adjust marketing strategies.

A comprehensive study of data intelligence in the context of big data analytics

Modern systems like the Internet of Things, cloud computing, and sensor networks generate a huge data archive. The knowledge extraction from these huge archived data requires modified approaches in algorithm design techniques. The field of study in which analysis of such huge data is carried out is called big data analytics, which helps to optimize the performance with reduced cost and retrieves the information efficiently. The enhancement of traditional data analytics needs to modify to suit big data analytics because it may not manage huge amounts of data. The real thought is how to design the data mining algorithms suitable to handle big data analysis. This paper discusses data analytics at the initial level, to begin with, the insights about the analysis process for big data. Big data analytics have a current research edge in the knowledge extraction field. This paper highlights the challenges and problems associated with big data analysis and provide inner insights into several techniques and methods used.

Predicting the Vulnerability of Women to Intimate Partner Violence in South Africa: Evidence from Tree-based Machine Learning Techniques

Intimate partner violence (IPV) is a pervasive social challenge with severe health and demographic consequences. Global statistics indicate that more than a third of women have experienced IPV at some point in their lives. In South Africa, IPV is considered a significant contributor to the country's broader problem with violence and a leading cause of femicide. Consequently, IPV has been the major focus of legislation and research across different disciplines. The present article aims to contribute to the growing scholarly literature by predicting factors that are associated with the risk of experiencing IPV. We used the 2016 South African Demographic and Health Survey dataset and restricted our analysis to 1,816 ever-married women who had complete information on the variables that were used to generate IPV. Prior research has mainly used regression analysis to identify correlates of IPV; however, while regression analysis can test a priori specified effects, it cannot capture unspecified inter-relationship across factors. To address this limitation, we opted for machine learning methods, which identify hidden and complex patterns and relationships in the data. Our results indicate that the fear of the husband is the most critical factor in determining the experience of IPV. In other words, the risk of IPV in South Africa is associated more with the husband or partner's characteristics than the woman's. The models developed in this study can be used to develop interventions by different stakeholders such as social workers, policymakers, and or other interested partners.

Automated Detection of Electric Energy Consumption Load Profile Patterns

Load profiles of energy consumption from smart meters are becoming more and more available, and the amount of data to analyse is huge. In order to automate this analysis, the application of state-of-the-art data mining techniques for time series analysis is reviewed. In particular, the use of dynamic clustering techniques to obtain and visualise temporal patterns characterising the users of electrical energy is deeply studied. The performed review can be used as a guide for those interested in the automatic analysis and groups of behaviour detection within load profile databases. Additionally, a selection of dynamic clustering algorithms have been implemented and the performances compared using an available electric energy consumption load profile database. The results allow experts to easily evaluate how users consume energy, to assess trends and to predict future scenarios.

Attribute dependency data analysis for massive datasets by fuzzy transforms

We present a numerical attribute dependency method for massive datasets based on the concepts of direct and inverse fuzzy transform. In a previous work, we used these concepts for numerical attribute dependency in data analysis: Therein, the multi-dimensional inverse fuzzy transform was useful for approximating a regression function. Here we give an extension of this method in massive datasets because the previous method could not be applied due to the high memory size. Our method is proved on a large dataset formed from 402,678 census sections of the Italian regions provided by the Italian National Statistical Institute (ISTAT) in 2011. The results of comparative tests with the well-known methods of regression, called support vector regression and multilayer perceptron, show that the proposed algorithm has comparable performance with those obtained using these two methods. Moreover, the number of parameters requested in our method is minor with respect to those of the cited in the above two algorithms.

RS-HeRR: a rough set-based Hebbian rule reduction neuro-fuzzy system

Interpretabilty is one of the desired characteristics in various classification task. Rule-based system and fuzzy logic can be used for interpretation in classification. The main drawback of rule-based system is that it may contain large complex rules for classification and sometimes it becomes very difficult in interpretation. Rule reduction is also difficult for various reasons. Removing important rules may effect in classification accuracy. This paper proposes a hybrid fuzzy-rough set approach named RS-HeRR for the generation of effective, interpretable and compact rule set. It combines a powerful rule generation and reduction fuzzy system, called Hebbian-based rule reduction algorithm (HeRR) and a novel rough-set-based attribute selection algorithm for rule reduction. The proposed hybridization leverages upon rule reduction through reduction in partial dependency as well as improvement in system performance to significantly reduce the problem of redundancy in HeRR, even while providing similar or better accuracy. RS-HeRR demonstrates these characteristics repeatedly over four diverse practical classification problems, such as diabetes identification, urban water treatment monitoring, sonar target classification, and detection of ovarian cancer. It also demonstrates excellent performance for highly biased datasets. In addition, it competes very well with established non-fuzzy classifiers and outperforms state-of-the-art methods that use rough sets for rule reduction in fuzzy systems.

Psychosocial Risks Assessment in Cryopreservation Laboratories

Background

Psychosocial risks are increasingly a type of risk analyzed in organizations beyond chemical, physical, and biological risks. To this type of risk, a greater attention has been given following the update of ISO 9001: 2015, more precisely the requirement 7.1.4 for the process operation environment. The update of this normative reference was intended to approximate OHSAS 18001: 2007 reference updated in 2018 with the publication of ISO 45001. Thus, the organizations are increasingly committed to achieving and demonstrating good occupational health and safety performance.

Methods

The aim of this study was to characterize the psychosocial risks in a cryopreservation laboratory and to develop a predictive model for psychosocial risk management. The methodology followed to collect the information was the inquiry by questionnaire that was applied to a sample comprising 200 employees.

Results

The results show that most of the respondents are aware of the psychosocial risks, identifying interpersonal relationships and emotional feelings as the main factors that lead to this type of risks. Furthermore, terms such as lack of resources, working hours, lab equipment, stress, and precariousness show strong correlation with psychosocial risks. The model presented in this study, based on artificial neural networks, exhibited good performance in the prediction of the psychosocial risks.

Conclusion

This work presents the development of an intelligent system that allows identifying the weaknesses of the organization and contributing to the enhancement of the psychosocial risks management.

Social media mining for smart cities and smart villages research

The imperative of well-being and improved quality of life in smart cities context can only be attained if the smart services, so central to the concept of smart cities, correspond with the needs, expectations and skills of cities’ inhabitants. Considering that social media generate and/or open real-time entry points to vast amounts of data pertinent to well-being and quality of life, such as citizens’ expectations, opinions, as well as to recent developments related to regulatory frameworks, debates, political decisions and policymaking, the big question is how to exploit the potential inherent in social media and use it to enhance the value added smart cities generate. Social mining is traditionally understood as the process of representing, analyzing, and extracting actionable patterns and trends from raw social media data. In the context of smart cities, this special issue focuses on how social media data, also potentially combined with other data, can be used to optimize the efficiency of city operations and services, and thereby contribute more efficiently to citizens’ well-being and quality of life.

Embedding Learning with Triple Trustiness on Noisy Knowledge Graph

Embedding learning on knowledge graphs (KGs) aims to encode all entities and relationships into a continuous vector space, which provides an effective and flexible method to implement downstream knowledge-driven artificial intelligence (AI) and natural language processing (NLP) tasks. Since KG construction usually involves automatic mechanisms with less human supervision, it inevitably brings in plenty of noises to KGs. However, most conventional KG embedding approaches inappropriately assume that all facts in existing KGs are completely correct and ignore noise issues, which brings about potentially serious errors. To address this issue, in this paper we propose a novel approach to learn embeddings with triple trustiness on KGs, which takes possible noises into consideration. Specifically, we calculate the trustiness value of triples according to the rich and relatively reliable information from large amounts of entity type instances and entity descriptions in KGs. In addition, we present a cross-entropy based loss function for model optimization. In experiments, we evaluate our models on KG noise detection, KG completion and classification. Through extensive experiments on three datasets, we demonstrate that our proposed model can learn better embeddings than all baselines on noisy KGs.

Predicting skilled delivery service use in Ethiopia: dual application of logistic regression and machine learning algorithms

Background

Skilled assistance during childbirth is essential to reduce maternal deaths. However, in Ethiopia, which is among the six countries contributing to more than half of the global maternal deaths, the coverage of births attended by skilled health personnel remains very low. The aim of this study was to identify determinants and develop a predictive model for skilled delivery service use in Ethiopia by applying logistic regression and machine-learning techniques.

Methods

Data from the 2016 Ethiopian Demographic and Health Survey (EDHS) was used for this study. Statistical Package for Social Sciences (SPSS) and Waikato Environment for Knowledge Analysis (WEKA) tools were used for logistic regression and model building respectively. Classification algorithms namely J48, Naïve Bayes, Support Vector Machine (SVM), and Artificial Neural Network (ANN) were used for model development. The validation of the predictive models was assessed using accuracy, sensitivity, specificity, and area under Receiver Operating Characteristics (ROC) curve.

Results

Only 27.7% women received skilled delivery assistance in Ethiopia. First antenatal care (ANC) [AOR = 1.83, 95% CI (1.24–2.69)], birth order [AOR = 0.22, 95% CI (0.11–0.46)], television ownership [AOR = 6.83, 95% CI (2.52–18.52)], contraceptive use [AOR = 1.92, 95% CI (1.26–2.97)], cost needed for healthcare [AOR = 2.17, 95% CI (1.47–3.21)], age at first birth [AOR = 1.96, 95% CI (1.31–2.94)], and age at first sex [AOR = 2.72, 95% CI (1.55–4.76)] were determinants for utilizing skilled delivery services during the childbirth. Predictive models were developed and the J48 model had superior predictive accuracy (98%), sensitivity (96%), specificity (99%) and, the area under ROC (98%).

Conclusions

First ANC and contraceptive uses were among the determinants of utilization of skilled delivery services. A predictive model was developed to forecast the likelihood of a pregnant woman seeking skilled delivery assistance; therefore, the predictive model can help to decide targeted interventions for a pregnant woman to ensure skilled assistance at childbirth. The model developed through the J48 algorithm has better predictive accuracy. Web-based application can be build based on results of this study.

Review of Soft Computing Models in Design and Control of Rotating Electrical Machines

Rotating electrical machines are electromechanical energy converters with a fundamental impact on the production and conversion of energy. Novelty and advancement in the control and high-performance design of these machines are of interest in energy management. Soft computing methods are known as the essential tools that significantly improve the performance of rotating electrical machines in both aspects of control and design. From this perspective, a wide range of energy conversion systems such as generators, high-performance electric engines, and electric vehicles, are highly reliant on the advancement of soft computing techniques used in rotating electrical machines. This article presents the-state-of-the-art of soft computing techniques and their applications, which have greatly influenced the progression of this significant realm of energy. Through a novel taxonomy of systems and applications, the most critical advancements in the field are reviewed for providing an insight into the future of control and design of rotating electrical machines.

Water demand forecasting: review of soft computing methods

Demand forecasting plays a vital role in resource management for governments and private companies. Considering the scarcity of water and its inherent constraints, demand management and forecasting in this domain are critically important. Several soft computing techniques have been developed over the last few decades for water demand forecasting. This study focuses on soft computing methods of water consumption forecasting published between 2005 and 2015. These methods include artificial neural networks (ANNs), fuzzy and neuro-fuzzy models, support vector machines, metaheuristics, and system dynamics. Furthermore, it was discussed that while in short-term forecasting, ANNs have been superior in many cases, but it is still very difficult to pick a single method as the overall best. According to the literature, various methods and their hybrids are applied to water demand forecasting. However, it seems soft computing has a lot more to contribute to water demand forecasting. These contribution areas include, but are not limited, to various ANN architectures, unsupervised methods, deep learning, various metaheuristics, and ensemble methods. Moreover, it is found that soft computing methods are mainly used for short-term demand forecasting.

Monitoring and Discovery for Self-Organized Network Management in Virtualized and Software Defined Networks

This paper presents the Monitoring and Discovery Framework of the Self-Organized Network Management in Virtualized and Software Defined Networks SELFNET project. This design takes into account the scalability and flexibility requirements needed by 5G infrastructures. In this context, the present framework focuses on gathering and storing the information (low-level metrics) related to physical and virtual devices, cloud environments, flow metrics, SDN traffic and sensors. Similarly, it provides the monitoring data as a generic information source in order to allow the correlation and aggregation tasks. Our design enables the collection and storing of information provided by all the underlying SELFNET sublayers, including the dynamically onboarded and instantiated SDN/NFV Apps, also known as SELFNET sensors.

Determinants and development of a web-based child mortality prediction model in resource-limited settings: A data mining approach

BACKGROUND

Improving child health and reducing child mortality rate are key health priorities in developing countries. This study aimed to identify determinant sand develop, a web-based child mortality prediction model in Ethiopian local language using classification data mining algorithm.

METHODS

Decision tree (using J48 algorithm) and rule induction (using PART algorithm) techniques were applied on 11,654 records of Ethiopian demographic and health survey data. Waikato Environment for Knowledge Analysis (WEKA) for windows version 3.6.8 was used to develop optimal models. 8157 (70%) records were randomly allocated to training group for model building while; the remaining 3496 (30%) records were allocated as the test group for model validation. The validation of the model was assessed using accuracy, sensitivity, specificity and area under Receiver Operating Characteristics (ROC) curve. Using Statistical Package for Social Sciences (SPSS) version 20.0; logistic regressions and Odds Ratio (OR) with 95% Confidence Interval (CI) was used to identify determinants of child mortality.

RESULTS

The child mortality rate was 72 deaths per 1000 live births. Breast-feeding (AOR= 1.46, (95% CI [1.22. 1.75]), maternal education (AOR= 1.40, 95% CI [1.11, 1.81]), family planning (AOR= 1.21, [1.08, 1.43]), preceding birth interval (AOR= 4.90, [2.94, 8.15]), presence of diarrhea (AOR= 1.54, 95% CI [1.32, 1.66]), father's education (AOR= 1.4, 95% CI [1.04, 1.78]), low birth weight (AOR= 1.2, 95% CI [0.98, 1.51]) and, age of the mother at first birth (AOR= 1.42, [1.01-1.89]) were found to be determinants for child mortality. The J48 model had better performance, accuracy (94.3%), sensitivity (93.8%), specificity (94.3%), Positive Predictive Value (PPV) (92.2%), Negative Predictive Value (NPV) (94.5%) and, the area under ROC (94.8%). Subsequent to developing an optimal prediction model, we relied on this model to develop a web-based application system for child mortality prediction.

CONCLUSION

In this study, nearly accurate results were obtained by employing decision tree and rule induction techniques. Determinants are identified and a web-based child mortality prediction model in Ethiopian local language is developed. Thus, the result obtained could support child health intervention programs in Ethiopia where trained human resource for health is limited. Advanced classification algorithms need to be tested to come up with optimal models.

A novel fuzzy approach for automatic Brunnstrom stage classification using surface electromyography

Clinical assessment plays a major role in post-stroke rehabilitation programs for evaluating impairment level and tracking recovery progress. Conventionally, this process is manually performed by clinicians using chart-based ordinal scales which can be both subjective and inefficient. In this paper, a novel approach based on fuzzy logic is proposed which automatically evaluates stroke patients' impairment level using single-channel surface electromyography (sEMG) signals and generates objective classification results based on the widely used Brunnstrom stages of recovery. The correlation between stroke-induced motor impairment and sEMG features on both time and frequency domain is investigated, and a specifically designed fuzzy kernel classifier based on geometrically unconstrained membership function is introduced in the study to tackle the challenges in discriminating data classes with complex separating surfaces. Experiments using sEMG data collected from stroke patients have been carried out to examine the validity and feasibility of the proposed method. In order to ensure the generalization capability of the classifier, a cross-validation test has been performed. The results, verified using the evaluation decisions provided by an expert panel, have reached a rate of success of the 92.47%. The proposed fuzzy classifier is also compared with other pattern recognition techniques to demonstrate its superior performance in this application.

Identifying the perspectives for sustainability enhancement

Purpose

Integrating sustainability strategies with business processes is the most challenging task for industry professionals due to the lack of a proper understanding of sustainability concepts. At the same time, a lack of proper guidance restricts them from pursuing such activities. As far as the aspects of implementation are concerned, it is very tough to analyse and pick up key points to start with. The purpose of this paper is to utilize a text mining approach to analyse qualitative data and identify the critical issues for implementing sustainability in the manufacturing sector by focussing on turning processes based on the survey responses of researchers and industry professionals.

Design/methodology/approach

An integrated method employing principal component analysis (PCA) and the k-means clustering algorithm has been applied to extract useful information from a set of various suggestions provided by both the groups surveyed. The textual data has also been visualized using word clouds and, thus, it has been compared with the results of the text mining approach.

Findings

The results of the study indicate the importance of the role of government organizations and the need for a skilled workforce, which are crucial for enhancing aspects of sustainability in the manufacturing sector, as supported by both researchers and industry professionals. Besides this, researchers have highlighted the need to focus more on environmentally related issues, whereas industry professionals have raised performance-related issues.

Practical implications

The findings of the study present the important concerns of both the groups towards sustainability initiatives and, thus, will help to enhance the understanding of the underlying possibilities of negotiating jointly to enhance the performance of machining processes.

Originality/value

The novelty of this paper lies in its identification of important initiatives that are having a direct impact on the sustainable aspects of the machining process, based on the views of researchers and industry professionals.

Efficient Retrieval Technique for Microarray Gene Expression

The DNA mciroarray gene data is in the expression levels of thousands of genes for a small amount of samples. From the microarray gene data, the process of extracting the required knowledge remains an open challenge. Acquiring knowledge is the intricacy in such types of gene data, though number of researches is arising in order to acquire information from these gene data. In order to retrieve the required information, gene classification is vital; however, the task is complex because of the data characteristics, high dimensionality and smaller sample size. Initially, the dimensionality diminution process is carried out in order to shrink the microarray data without losing information with the aid of LPP and PCA techniques and utilized for information retrieval. In this paper, we propose an effective gene retrieval technique based on LPP and PCA called LPCA. The technique like LPP and PCA is chosen for the dimensionality reduction for efficient retrieval of microarray gene data. An application of microarray gene data is included with classification by SVM. SVM is trained by the dimensionality reduced gene data for effective classification. A comparative study is made with these dimensionality reduction techniques.

ANALYZING MINING PATTERNS USING FUZZY ART AND SOFT REGRESSION

Data mining is widely used to solve real-world problems in engineering, science and business. Usually the results from data mining obtained through the traditional approaches are not interpretable in the real life scenario. This paper suggests an approach for logical interpretations of the clustered data. Our approach involves using fuzzy ART technique for clustering the data and then applying the soft regression technique for interpreting the results of the clustering. The proposed model provides better analysis of data for describing overlapping clusters. We used our model to analyze patterns in the advances data of a public sector bank. Analyses and experiments show the effectiveness of the proposed method.

NEURAL NETWORK-BASED INTERPOLATION AND EXTRAPOLATION OF WIND TUNNEL TEST DATA

There is an increasing need to apply emerging technologies in knowledge-based management system. This proposed system captures the knowledge of experts and the knowledge acquired for designing a new systems. Wind Tunnel Test data of missiles has been taken into consideration for knowledge management. Interpolation and Extrapolation of missile test data has been attempted using neural network technique General Regression Neural Networks training algorithm. The results produced by neural network training methodologies are validated with the existing test data. The knowledge extraction using neural network is found suitable for interpolation and to some extent for extrapolation, thereby reducing the cost as well as number of test runs in Wind Tunnel Test.

BUILDING CUSTOMER MODELS FROM BUSINESS DATA: AN AUTOMATIC APPROACH BASED ON FUZZY CLUSTERING AND MACHINE LEARNING

Data mining (DM) is a new emerging discipline that aims to extract knowledge from data using several techniques. DM turned out to be useful in business where the data describing the customers and their transactions is in the order of terabytes. In this paper, we propose an approach for building customer models (said also profiles in the literature) from business data. Our approach is three-step. In the first step, we use fuzzy clustering to categorize customers, i.e., determine groups of customers. A key feature is that the number of groups (or clusters) is computed automatically from data using the partition entropy as a validity criteria. In the second step, we proceed to a dimensionality reduction which aims at keeping for each group of customers only the most informative attributes. For this, we define the information loss to quantify the information degree of an attribute. Hence, and as a result to this second step, we obtain groups of customers each described by a distinct set of attributes. In the third and final step, we use backpropagation neural networks to extract useful knowledge from these groups. Experimental results on real-world data sets reveal a good performance of our approach and should simulate future research.

Artificial odor discrimination system using electronic nose and neural networks for the identification of urinary tract infection

Current clinical diagnostics are based on biochemical, immunological, or microbiological methods. However, these methods are operator dependent, time-consuming, expensive, and require special skills, and are therefore, not suitable for point-of-care testing. Recent developments in gas-sensing technology and pattern recognition methods make electronic nose technology an interesting alternative for medical point-of-care devices. An electronic nose has been used to detect urinary tract infection from 45 suspected cases that were sent for analysis in a U.K. Public Health Registry. These samples were analyzed by incubation in a volatile generation test tube system for 4-5 h. Two issues are being addressed, including the implementation of an advanced neural network, based on a modified expectation maximization scheme that incorporates a dynamic structure methodology and the concept of a fusion of multiple classifiers dedicated to specific feature parameters. This study has shown the potential for early detection of microbial contaminants in urine samples using electronic nose technology.