José D. Martín

BELM: Bayesian Extreme Learning Machine

The theory of extreme learning machine (ELM) has become very popular on the last few years. ELM is a new approach for learning the parameters of the hidden layers of a multilayer neural network (as the multilayer perceptron or the radial basis function neural network). Its main advantage is the lower computational cost, which is especially relevant when dealing with many patterns defined in a high-dimensional space. This brief proposes a Bayesian approach to ELM, which presents some advantages over other approaches: it allows the introduction of a priori knowledge; obtains the confidence intervals (CIs) without the need of applying methods that are computationally intensive, e.g., bootstrap; and presents high generalization capabilities. Bayesian ELM is benchmarked against classical ELM in several artificial and real datasets that are widely used for the evaluation of machine learning algorithms. Achieved results show that the proposed approach produces a competitive accuracy with some additional advantages, namely, automatic production of CIs, reduction of probability of model overfitting, and use of a priori knowledge.

Neural networks for animal science applications: Two case studies

Abstract Artificial neural networks have shown to be a powerful tool for system modelling in a wide range of applications. In this paper, we focus on neural network applications to intelligent data analysis in the field of animal science. Two classical applications of neural networks are proposed: time series prediction and clustering. The first task is related to the prediction of weekly milk production in goat flocks, which includes a knowledge discovery stage in order to analyse the relative relevance of the different variables. The second task is the clustering of goat flocks; it is used to analyse different livestock surveys by using self-organizing maps and the adaptive resonance theory, thus obtaining a qualitative knowledge from these surveys. Achieved results show the usefulness of neural networks in two animal science applications.

An integrated framework for risk profiling of breast cancer patients following surgery

OBJECTIVE An integrated decision support framework is proposed for clinical oncologists making prognostic assessments of patients with operable breast cancer. The framework may be delivered over a web interface. It comprises a triangulation of prognostic modelling, visualisation of historical patient data and an explanatory facility to interpret risk group assignments using empirically derived Boolean rules expressed directly in clinical terms. METHODS AND MATERIALS The prognostic inferences in the interface are validated in a multicentre longitudinal cohort study by modelling retrospective data from 917 patients recruited at Christie Hospital, Wilmslow between 1983 and 1989 and predicting for 931 patients recruited in the same centre during 1990-1993. There were also 291 patients recruited between 1984 and 1998 at the Clatterbridge Centre for Oncology and the Linda McCartney Centre, Liverpool, UK. RESULTS AND CONCLUSIONS There are three novel contributions relating this paper to breast cancer cases. First, the widely used Nottingham prognostic index (NPI) is enhanced with additional clinical features from which prognostic assessments can be made more specific for patients in need of adjuvant treatment. This is shown with a cross matching of the NPI and a new prognostic index which also provides a two-dimensional visualisation of the complete patient database by risk of negative outcome. Second, a principled rule-extraction method, orthogonal search rule extraction, generates readily interpretable explanations of risk group allocations derived from a partial logistic artificial neural network with automatic relevance determination (PLANN-ARD). Third, 95% confidence intervals for individual predictions of survival are obtained by Monte Carlo sampling from the PLANN-ARD model.

Feature selection using ROC curves on classification problems

Feature Selection (FS) is one of the key stages in classification problems. This paper proposes the use of the area under Receiver Operator Characteristic curves to measure the individual importance of every input as well as a method to discover the variables that yield a statistically significant improvement in the discrimination power of the classification model.

A Novel Semi-Supervised Methodology for Extracting Tumor Type-Specific MRS Sources in Human Brain Data

Background The clinical investigation of human brain tumors often starts with a non-invasive imaging study, providing information about the tumor extent and location, but little insight into the biochemistry of the analyzed tissue. Magnetic Resonance Spectroscopy can complement imaging by supplying a metabolic fingerprint of the tissue. This study analyzes single-voxel magnetic resonance spectra, which represent signal information in the frequency domain. Given that a single voxel may contain a heterogeneous mix of tissues, signal source identification is a relevant challenge for the problem of tumor type classification from the spectroscopic signal. Methodology/Principal Findings Non-negative matrix factorization techniques have recently shown their potential for the identification of meaningful sources from brain tissue spectroscopy data. In this study, we use a convex variant of these methods that is capable of handling negatively-valued data and generating sources that can be interpreted as tumor class prototypes. A novel approach to convex non-negative matrix factorization is proposed, in which prior knowledge about class information is utilized in model optimization. Class-specific information is integrated into this semi-supervised process by setting the metric of a latent variable space where the matrix factorization is carried out. The reported experimental study comprises 196 cases from different tumor types drawn from two international, multi-center databases. The results indicate that the proposed approach outperforms a purely unsupervised process by achieving near perfect correlation of the extracted sources with the mean spectra of the tumor types. It also improves tissue type classification. Conclusions/Significance We show that source extraction by unsupervised matrix factorization benefits from the integration of the available class information, so operating in a semi-supervised learning manner, for discriminative source identification and brain tumor labeling from single-voxel spectroscopy data. We are confident that the proposed methodology has wider applicability for biomedical signal processing.

Infrared thermography is useful for ruling out fractures in paediatric emergencies

AbstractMusculoskeletal injuries are a leading cause of paediatric injuries and emergency department visits in Western countries. Diagnosis usually involves radiography, but this exposes children without fractures to unnecessary ionising radiation. We explored whether infrared thermography could provide a viable alternative in trauma cases. We compared radiography and thermal images of 133 children who had been diagnosed with a trauma injury in the emergency unit of a Spanish hospital. As well as the thermal variables in the literature, we introduced a new quantifier variable, the size of the lesion. Decision tree models were built to assess the technique’s accuracy in diagnosing whether a bone had been fractured or not. Infrared thermography had a sensitivity of 0.91, a specificity of 0.88 and a negative predictive value of 0.95. The new lesion size variable introduced appeared to be of main importance to the discriminatory power of the method. Conclusion: The high negative predictive value of infrared thermography suggests that it is a promising method for ruling out fractures.

Cloud detection for CHRIS/Proba hyperspectral images

Accurate and automatic detection of clouds in satellite scenes is a key issue for a wide range of remote sensing applications. With no accurate cloud masking, undetected clouds are one of the most significant source of error in both sea and land cover biophysical parameter retrieval. Sensors with spectral channels beyond 1 um have demonstrated good capabilities to perform cloud masking. This spectral range can not be exploited by recently developed hyperspectral sensors that work in the spectral range between 400- 1000 nm. However, one can take advantage of their high number of channels and spectral resolution to increase the cloud detection accuracy, and to describe properly the detected clouds (cloud type, height, subpixel coverage, could shadows, etc.) In this paper, we present a methodology for cloud detection that could be used by sensors working in the VNIR range. First, physically-inspired features are extracted (TOA reflectance and their spectral derivatives, atmospheric oxygen and water vapour absorptions, etc). Second, growing maps are built from cloud-like pixels to select regions which potentially could contain clouds. Then, an unsupervised clustering algorithm is applied in these regions using all extracted features. The obtained clusters are labeled into geo-physical classes taking into account the spectral signature of the cluster centers. Finally, an spectral unmixing algorithm is applied to the segmented image in order to obtain an abundance map of the cloud content in the cloud pixels. As a direct consequence of the detection scheme, the proposed system is capable to yield probabilistic outputs on cloud detected pixels in the image, rather than flags. Performance of the proposed algorithm is tested on six CHRIS/Proba Mode 1 images, which presents a spatial resolution of 32 m, 62 spectral bands with 6-20 nm bandwidth, and multiangularity.

The use of neural networks for predicting the result of endoscopic treatment for vesico‐ureteric reflux

To create an artificial neural network (ANN) to aid in predicting the results of endoscopic treatment for vesico‐ureteric reflux (VUR).

Neural networks as effective techniques in clinical management of patients: some case studies

In this paper, we present four examples of effective implementation of neural systems in the daily clinical practice. There are two main goals in this work; the first one is to show that neural networks are especially well-suited tools for solving different kind of medical/pharmaceutical problems, given the complex input output relationships and the few a priori knowledge about data distribution and variable relations. The second goal is to develop specific software applications, which enclose complex mathematical models, to clinicians; thus, the use of such models as decision support systems is facilitated. Four important pharmaceutical problems are considered in this study: identification of patients with potential risk of postchemotherapy emesis, classification of patients depending on their risk of digoxin intoxication, prediction of cyclosporine A through concentration and prediction of erythropoietin blood concentrations. The Multilayer Perceptron in classification problems and dynamic neural networks, such as the Elman recurrent neural network and the Finite Impulse Response neural network in prediction problems, have been used. Moreover, network ensembles of different kind of networks have been taken into account. Results show that neural networks are suitable tools for medical classification and prediction tasks, outperforming the mostly used methods in these problems (logistic regression and multivariate analysis).

Towards interpretable classifiers with blind signal separation

Blind signal separation (BSS) is a powerful tool to open-up complex signals into component sources that are often interpretable. However, BSS methods are generally unsupervised, therefore the assignment of class membership from the elements of the mixing matrix may be sub-optimal. This paper proposes a three-stage approach using Fisher information metric to define a natural metric for the data, from which a Euclidean approximation can then be used to drive BSS. Results with synthetic data models of real-world high-dimensional data show that the classification accuracy of the method is good for challenging problems, while retaining interpretability.