eXiT*CBR: A framework for case-based medical diagnosis development and experimentation

Objective

Medical applications have special features (interpretation of results in medical metrics, experiment reproducibility and dealing with complex data) that require the development of particular tools. The eXiT*CBR framework is proposed to support the development of and experimentation with new case-based reasoning (CBR) systems for medical diagnosis.

Method

Our framework offers a modular, heterogeneous environment that combines different CBR techniques for different application requirements. The graphical user interface allows easy navigation through a set of experiments that are pre-visualized as plots (receiver operator characteristics (ROC) and accuracy curves). This user-friendly navigation allows easy analysis and replication of experiments. Used as a plug-in on the same interface, eXiT*CBR can work with any data mining technique such as determining feature relevance.

Results

The results show that eXiT*CBR is a user-friendly tool that facilitates medical users to utilize CBR methods to determine diagnoses in the field of breast cancer, dealing with different patterns implicit in the data.

Conclusions

Although several tools have been developed to facilitate the rapid construction of prototypes, none of them has taken into account the particularities of medical applications as an appropriate interface to medical users. eXiT*CBR aims to fill this gap. It uses CBR methods and common medical visualization tools, such as ROC plots, that facilitate the interpretation of the results. The navigation capabilities of this tool allow the tuning of different CBR parameters using experimental results. In addition, the tool allows experiment reproducibility.     

Error-correction learning for artificial neural networks using the Bayesian paradigm. Application to automated medical diagnosis

Automated medical diagnosis models are now ubiquitous, and research for developing new ones is constantly growing. They play an important role in medical decision-making, helping physicians to provide a fast and accurate diagnosis. Due to their adaptive learning and nonlinear mapping properties, the artificial neural networks are widely used to support the human decision capabilities, avoiding variability in practice and errors based on lack of experience. Among the most common learning approaches, one can mention either the classical back-propagation algorithm based on the partial derivatives of the error function with respect to the weights, or the Bayesian learning method based on posterior probability distribution of weights, given training data. This paper proposes a novel training technique gathering together the error-correction learning, the posterior probability distribution of weights given the error function, and the Goodman–Kruskal Gamma rank correlation to assembly them in a Bayesian learning strategy. This study had two main purposes; firstly, to develop a novel learning technique based on both the Bayesian paradigm and the error back-propagation, and secondly, to assess its effectiveness. The proposed model performance is compared with those obtained by traditional machine learning algorithms using real-life breast and lung cancer, diabetes, and heart attack medical databases. Overall, the statistical comparison results indicate that the novel learning approach outperforms the conventional techniques in almost all respects.     

How to use contextual knowledge in medical case-based reasoning systems: a survey on very recent trends

Objectives

This paper aims at systematizing the ways in which the contextual knowledge embedded in the case library can support decision making, within case-based reasoning (CBR) systems. In particular, CBR applications to the medical domain are considered.

Methods and material

After a quick survey on the definition and on the role of context in artificial intelligence research, we have focused on CBR, with a particular emphasis on medical applications. In this field, we have identified a number of very recent contributions, which strongly recognize context per se as a major knowledge source. These contributions propose to maintain and to rely on contextual information, in order to support human reasoning in different fashions.

Results

We have distinguished three main directions in which contextual knowledge can be resorted to, in order to optimize physicians’ decision making. Such directions can be summarized as follows: (1) to reduce the search space in the case retrieval step; (2) to maintain the overall knowledge content always valid and up to date, and (3) to adapt knowledge application and reasoning to local/personal constraints. We have also properly categorized the surveyed works within these three clusters, and identified the most significant ones, able to exploit contextual knowledge along more than one direction.

Conclusions

Innovative applications of the contextual knowledge recorded in the case library, described and systematized in this paper, can trace promising research directions for the future.     

Boosting backpropagation algorithm by stimulus-sampling: Application in computer-aided medical diagnosis

Neural networks (NNs), in general, and multi-layer perceptron (MLP), in particular, represent one of the most efficient classifiers among the machine learning (ML) algorithms. Inspired by the stimulus-sampling paradigm, it is plausible to assume that the association of stimuli with the neurons in the output layer of a MLP can increase its performance. The stimulus-sampling process is assumed memoryless (Markovian), in the sense that the choice of a particular stimulus at a certain step, conditioned by the whole prior evolution of the learning process, depends only on the network’s answer at the previous step. This paper proposes a novel learning technique, by enhancing the standard backpropagation algorithm performance with the aid of a stimulus-sampling procedure applied to the output neurons. The network uses the observable behavior that varies throughout the training process by stimulating the correct answers through corresponding rewards/penalties assigned to the output neurons. The proposed model has been applied in computer-aided medical diagnosis using five real-life breast cancer, colon cancer, diabetes, thyroid, and fetal heartbeat databases. The statistical comparison to well-established ML algorithms proved beyond doubt its efficiency and robustness.     

Using a fuzzy controller optimized by a genetic algorithm to regulate blood glucose level in type 1 diabetes

In this paper a closed-loop control algorithm for blood glucose regulation in type 1 diabetic patients is proposed by using the Mamdani-type fuzzy method. Because of the presence of high-pass proportional derivatives in fuzzy designing, optimal values are applied for two inputs and one output membership functions in order to prevent the fluctuations due to derivatives in fuzzy design. Therefore, 19 values which are related to membership functions of the two inputs and one output are obtained by using a genetic algorithm (GA). The new model, termed the Augmented Minimal Model (AMM), is used in simulations. This controller is capable of stabilizing the blood glucose concentration at a normoglycaemic level of 90?mg dl^?1. The operation of the controller under various situations including multiple meal disturbances, and noise due to inaccurate effects of measuring blood glucose level are considered. Uncertainties in the meal disturbance function and variations of model parameters were also taken into consideration in simulations and the controller was found to be robust to such uncertainties.     

Automatic detection of solitary lung nodules using quality threshold clustering,genetic algorithm and diversity index

ObjectiveThe present work has the objective of developing an automatic methodology for the detection of lung nodules.MethodologyThe proposed methodology is based on image processing and pattern recognition techniques and can be summarized in three stages. In the first stage, the extraction and reconstruction of the pulmonary parenchyma is carried out and then enhanced to highlight its structures. In the second stage, nodule candidates are segmented. Finally, in the third stage, shape and texture features are extracted, selected and then classified using a support vector machine.ResultsIn the testing stage, with 140 new exams from the Lung Image Database Consortium image collection, 80% of which are for training and 20% are for testing, good results were achieved, as indicated by a sensitivity of 85.91%, a specificity of 97.70% and an accuracy of 97.55%, with a false positive rate of 1.82 per exam and 0.008 per slice and an area under the free response operating characteristic of 0.8062.ConclusionLung cancer presents the highest mortality rate in addition to one of the smallest survival rates after diagnosis. An early diagnosis considerably increases the survival chance of patients. The methodology proposed herein contributes to this diagnosis by being a useful tool for specialists who are attempting to detect nodules.     

Pap smear diagnosis using a hybrid intelligent scheme focusing on genetic algorithm based feature selection and nearest neighbor classification

The term pap-smear refers to samples of human cells stained by the so-called Papanicolaou method. The purpose of the Papanicolaou method is to diagnose pre-cancerous cell changes before they progress to invasive carcinoma. In this paper a metaheuristic algorithm is proposed in order to classify the cells. Two databases are used, constructed in different times by expert MDs, consisting of 917 and 500 images of pap smear cells, respectively. Each cell is described by 20 numerical features, and the cells fall into 7 classes but a minimal requirement is to separate normal from abnormal cells, which is a 2 class problem. For finding the best possible performing feature subset selection problem, an effective genetic algorithm scheme is proposed. This algorithmic scheme is combined with a number of nearest neighbor based classifiers. Results show that classification accuracy generally outperforms other previously applied intelligent approaches.     

GMDH-based feature ranking and selection for improved classification of medical data

Medical applications are often characterized by a large number of disease markers and a relatively small number of data records. We demonstrate that complete feature ranking followed by selection can lead to appreciable reductions in data dimensionality, with significant improvements in the implementation and performance of classifiers for medical diagnosis. We describe a novel approach for ranking all features according to their predictive quality using properties unique to learning algorithms based on the group method of data handling (GMDH). An abductive network training algorithm is repeatedly used to select groups of optimum predictors from the feature set at gradually increasing levels of model complexity specified by the user. Groups selected earlier are better predictors. The process is then repeated to rank features within individual groups. The resulting full feature ranking can be used to determine the optimum feature subset by starting at the top of the list and progressively including more features until the classification error rate on an out-of-sample evaluation set starts to increase due to overfitting. The approach is demonstrated on two medical diagnosis datasets (breast cancer and heart disease) and comparisons are made with other feature ranking and selection methods. Receiver operating characteristics (ROC) analysis is used to compare classifier performance. At default model complexity, dimensionality reduction of 22 and 54% could be achieved for the breast cancer and heart disease data, respectively, leading to improvements in the overall classification performance. For both datasets, considerable dimensionality reduction introduced no significant reduction in the area under the ROC curve. GMDH-based feature selection results have also proved effective with neural network classifiers.     

Mathematical modeling and application of genetic algorithm to parameter estimation in signal transduction: trafficking and promiscuous coupling of G-protein coupled receptors

G-protein-coupled receptors (GPCRs) constitute a large and diverse family of proteins whose primary function is to transduce extracellular stimuli into intracellular signals. These receptors play a critical role in signal transduction, and are among the most important pharmacological drug targets. Upon binding of extracellular ligands, these receptor molecules couple to one or several subtypes of G-protein which reside at the intracellular side of the plasma membrane to trigger intracellular signaling events. The question of how GPCRs select and activate a single or multiple G-protein subtype(s) has been the topic of intense investigations. Evidence is also accumulating; however, that certain GPCRs can be internalized via lipid rafts and caveolae. In many cases, the mechanisms responsible for this still remain to be elucidated. In this work, we extend the mathematical model proposed by Chen et al. [Modelling of signalling via G-protein coupled receptors: pathway-dependent agonist potency and efficacy, Bull. Math. Biol. 65 (5) (2003) 933–958] to take into account internalization, recycling, degradation and synthesis of the receptors. In constructing the model, we assume that the receptors can exist in multiple conformational states allowing for a multiple effecter pathways. As data on kinetic reaction rates in the signalling processes measured in reliable in vivo and in vitro experiments is currently limited to a small number of known values. In this paper, we also apply a genetic algorithm (GA) to estimate the parameter values in our model.     

Genetic counseling in post-genomic era: Don’t pretend to know the meaning of a gene mutation if you don’t know

In this post-genomic era, more and more susceptibility loci of many possible genetic diseases are published. As our knowledge about these susceptibility loci is limited and partial, we should be very careful and responsible when patients seek genetic counseling about these possible genetic diseases. We should apply Confucius’s principle about knowledge and information to genetic conseling, and tell the truth to our patients about what we know and what we do not know. Like many other cancers, breast cancer is a very complicated, multifactorial disease; genetic factors, lifestyles and eating habits, environmental factors, and viral infections might be involved in breast cancer; hence, it is difficult to figure out the real etiology of breast cancer. It is not crystal clear that a person who carries mutations of the breast cancer 1, early onset and/or breast cancer 2, early onset genes would eventually get breast cancer in her/his lifetime. No person should undergo a preventive double mastectomy, unless we know the etiology of breast cancer someday.     

Induced tolerance and chimaerism in human fetuses using coelocentesis: a medical opportunity to avert genetic disease?

Coelocentesis offers a new opportunity for gaining access tothe coelomic cavity of human embryos from 28 days post-fertilization(42 days menstrual age). With this technique, cells can be extractedfrom the cavity for the genetic typing of embryos in early pregnancy.Coelocentesis may also offer a unique opportunity of inducingtolerance to foreign grafts and chimaerism in these human embryosby replacing donor cells into the coelomic cavity. This cavityappears to be closely associated with the fetal haemopoieticsystem. The optimal age to inject stem cells designed to producechimaerism may be at 5-6 weeks embryonic age, and these graftedcells may induce tolerance later in gestation. Two successivecoelocenteses would be needed, the first to extract fetal cellsto type the fetus, and a second within a few days to injectthe donor cells into the coelomic cavity. Alternatively, non-invasivemethods of diagnosis such as lower uterine pole extramembra-noussampling of fetal trophoblast, or the extraction of fetal cellsfrom maternal blood, could be combined with coelocentesis. Iftolerance and chimaerism can be established, repeated tissuegrafts could be carried out during fetal life and after birth,so that disorders caused by single or multiple gene defectsin the haemopoietic system and other organs may be corrected.