Search

International audience

The aim of the study was to compare open reduction with internal fixation (ORIF) and endoscopic open reduction with internal fixation (EORIF) of condylar fractures (CF) in adults in terms of reducing both needing of reoperation and/or facial nerve injury. An electronic search was undertaken (PubMed/MEDLINE, Web of Science, SCOPUS, and The Cochrane Library). The inclusion criteria were full text, published from their inception to June 2020, clinical trials, randomized or not, and retrospective studies, that compared ORIF and EORIF. The estimates of an intervention were expressed as the risk ratio (RR). From the 1338 articles found, 5 publications were included. There was no statistically significant difference between ORIF and EORIF regarding needing of reoperation (RR = 2.46, p = 0.42) or facial nerve injury (RR = 0.45, p = 0.14). Meta-analysis suggests that there is no difference between open reduction with internal fixation (ORIF) and endoscopic open reduction with internal fixation (EORIF) of condylar fractures (CF) regarding facial nerve injury risk or need for reoperation.

Standard classification algorithms ignore the time-related information contained in longitudinal data, as they do not consider the time indexes of the features’ different measurements. Accounting for temporal patterns may improve the algorithms’ performance, when applied to longitudinal data. Representing temporal patterns in the data itself has the advantage that those patterns are generic enough to be used with existing powerful classification algorithms, without requiring the design of new and more complex algorithms to exploit them. In this article, we propose 6 different types of constructed temporal features (3 of them being novel contributions), calculated from the values of the different feature measurements taken over time, and investigate whether adding those constructed temporal features to the original longitudinal dataset improves the classification model’s predictive accuracy. Our experiments involved 20 real-world longitudinal datasets created from a human-ageing study, and showed that the proposed approach of adding the constructed temporal features to the original feature set produced better classifiers overall.

Caloric restriction (CR) is the most studied pro-longevity intervention; however, a complete understanding of its underlying mechanisms remains elusive, and new research directions may emerge from the identification of novel CR-related genes and CR-related genetic features. This work used a Machine Learning (ML) approach to classify ageing-related genes as CR-related or NotCR-related using 9 different types of predictive features: PathDIP pathways, two types of features based on KEGG pathways, two types of Protein-Protein Interactions (PPI) features, Gene Ontology (GO) terms, Genotype-Tissue Expression (GTEx) expression features, Gene-Friends co-expression features and protein sequence descriptors. Our findings suggested that features biased towards curated knowledge (i.e. GO terms and biological pathways), had the greatest predictive power, while unbiased features (mainly gene expression and co-expression data) have the least predictive power. Moreover, a combination of all the feature types diminished the predictive power compared to predictions based on curated knowledge. Feature importance analysis on the two most predictive classifiers mostly corroborated existing knowledge and supported recent findings linking CR to the Nuclear Factor Erythroid 2-Related Factor 2 (NRF2) signalling pathway and G protein-coupled receptors (GPCR). We then used the two strongest combinations of feature type and ML algorithm to predict CR-relatedness among ageing-related genes currently lacking CR-related annotations in the data, resulting in a set of promising candidate CR-related genes (GOT2, GOT1, TSC1, CTH, GCLM, IRS2 and SESN2) whose predicted CR-relatedness remain to be validated in future wet-lab experiments.

Biologists very often use enrichment methods based on statistical hypothesis tests to identify gene properties that are significantly over-represented in a given set of genes of interest, by comparison with a ‘background’ set of genes. These enrichment methods, although based on rigorous statistical foundations, are not always the best single option to identify patterns in biological data. In many cases, one can also use classification algorithms from the machine-learning field. Unlike enrichment methods, classification algorithms are designed to maximize measures of predictive performance and are capable of analysing combinations of gene properties, instead of one property at a time. In practice, however, the majority of studies use either enrichment or classification methods (rather than both), and there is a lack of literature discussing the pros and cons of both types of method. The goal of this paper is to compare and contrast enrichment and classification methods, offering two contributions. First, we discuss the (to some extent complementary) advantages and disadvantages of both types of methods for identifying gene properties that discriminate between gene classes. Second, we provide a set of high-level recommendations for using enrichment and classification methods. Overall, by highlighting the strengths and the weaknesses of both types of methods we argue that both should be used in bioinformatics analyses.

Temporomandibular disorders (TMD) may present with diverse signs and symptoms, and one very significant is the limitation of mandibular movements. Additionally, the Helkimo indices allow for the reliable quantification of the signs and symptoms of TMD. The purpose of this study was to ascertain whether there are any correlations between the Helkimo indices and the maximal mandibular excursion capacity in a group of patients with TMD. Eighty patients (72 women and 8 men, mean age 33.6 years) with diagnosis of TMD according to the Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) were evaluated to obtain the Helkimo indices and their maximal mandibular excursion capacities. Normal or decreased maximum mandibular movements, i.e., opening, lateral and protrusion were compared with the anamnestic index, muscular pain symptoms, joint pain symptoms and the dysfunction index. Chi-squared analysis was used to compare normal and decreased movement capacities in terms of the Helkimo indices, and the muscle and temporomandibular joint (TMJ) pain. The statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) version 19.0. There were statistically significant differences in the clinical dysfunction index with the maximum opening (p = .011) and lateral movements (p = .024) but not with the maximum protrusion. There were no significant differences in the anamnestic index or the muscle pain and TMJ pain items of the clinical dysfunction index according to the mandibular excursions. The occurrences of limitations in the maximum opening and lateral movements are indications of greater TMD intensity.

Standard supervised machine learning methods often ignore the temporal information represented in longitudinal data, but that information can lead to more precise predictions in classification tasks. Data preprocessing techniques and classification algorithms can be adapted to cope directly with longitudinal data inputs, making use of temporal information such as the timeindex of features and previous measurements of the class variable. In this article, we propose two changes to the classification task of predicting age-related diseases in a real-world dataset created from the English Longitudinal Study of Ageing. First, we explore the addition of previous measurements of the class variable, and estimating the missing data in those added features using intermediate classifiers. Second, we propose a new splitfeature selection procedure for a random forest’s decision trees, which considers the candidate features’ time-indexes, in addition to the information gain ratio. Our experiments compared the proposed approaches to baseline approaches, in 3 prediction scenarios, varying the “time gap” for the prediction – how many years in advance the class (occurrence of an age-related disease) is predicted. The experiments were performed on 10 datasets varying the class variable, and showed that the proposed approaches increased the random forest’s predictive accuracy.

Ensembles of classifiers are a very popular type of method for performing classification, due to their usually high predictive accuracy. However, ensembles have two drawbacks. First, ensembles are usually considered a ‘black box’, non-interpretable type of classification model, mainly because typically there are a very large number of classifiers in the ensemble (and often each classifier in the ensemble is a black-box classifier by itself). This lack of interpretability is an important limitation in application domains where a model’s predictions should be carefully interpreted by users, like medicine, law, etc. Second, ensemble methods typically involve many hyper-parameters, and it is difficult for users to select the best settings for those hyper-parameters. In this work we propose an Evolutionary Algorithm (an Estimation of Distribution Algorithm) that addresses both these drawbacks. This algorithm optimizes the hyper-parameter settings of a small ensemble of 5 interpretable classifiers, which allows users to interpret each classifier. In our experiments, the ensembles learned by the proposed Evolutionary Algorithm achieved the same level of predictive accuracy as a well-known Random Forest ensemble, but with the benefit of learning interpretable models (unlike Random Forests).

This paper presents an experimental comparison among four automated machine learning (AutoML) methods for recommending the best classification algorithm for a given input dataset. Three of these methods are based on evolutionary algorithms (EAs), and the other is Auto-WEKA, a well-known AutoML method based on the combined algorithm selection and hyper-parameter optimisation (CASH) approach. The EA-based methods build classification algorithms from a single machine learning paradigm: either decision-tree induction, rule induction, or Bayesian network classification. Auto-WEKA combines algorithm selection and hyper-parameter optimisation to recommend classification algorithms from multiple paradigms. We performed controlled experiments where these four AutoML methods were given the same runtime limit for different values of this limit. In general, the difference in predictive accuracy of the three best AutoML methods was not statistically significant. However, the EA evolving decision-tree induction algorithms has the advantage of producing algorithms that generate interpretable classification models and that are more scalable to large datasets, by comparison with many algorithms from other learning paradigms that can be recommended by Auto-WEKA. We also observed that Auto-WEKA has shown meta-overfitting, a form of overfitting at the meta-learning level, rather than at the base-learning level.

Objective: To determine new body mass index (BMI) reference values to classify the nutritional status of children aged six to ten years old from the city of Montes Claros (state of Minas Gerais), Southeast Brazil. Methods: The sample consisted of 3,863 individuals from both genders. Body mass and height were measured to determine the BMI. We adopted the Lambda, Mu, and Sigma (LMS) method to obtain the cut-off points. After that, each stratum curve was smoothed using quartic polynomials by gender. Average interpolation was used to determine the biannual distribution values. We calculated the 3rd, 85th, and 95th centiles to classify underweight, overweight, and obesity, respectively, according to gender and age. Results: After tabulating the LMS parameters at biannual intervals by gender, we plotted a graphic with seven centiles of BMI distribution and calculated the new BMI parameters for children aged 6-10 years old from the city of Montes Claros. The cut-off values for underweight, overweight, and obesity classification were, respectively, 17.5, 25 and 30 kg/m2. Conclusions: For the studied children, the use of traditional BMI references may result in the overestimation of underweight and underestimation of overweight and obesity. Studies should be carried out with periodic updates, respecting the characteristics of each location in order to use BMI reference values to classify the nutritional status of children and adolescents.

In this work, we address the problem of feature selection for the classification task in hierarchical and sparse feature spaces, which characterise many real-world applications nowadays. A binary feature space is deemed hierarchical when its binary features are related via generalization-specialization relationships, and is considered sparse when in general the instances contain much fewer “positive” than “negative” feature values. In any given instance, a feature value is deemed positive (negative) when the property associated with the feature has been (has not been) observed for that instance. Although there are many methods for the traditional feature selection problem in the literature, the proper treatment to hierarchical feature structures is still a \ud challenge. Hence, we introduce a novel hierarchical feature selection method that follows the lazy learning paradigm – selecting a feature subset tailored for each instance in the test set. Our strategy prioritizes the selection of features with positive values, since they tend to be more informative – the presence of a relatively rare property is usually a piece of more relevant information than the absence of that property. Experiments on different application domains have shown that the proposed method outperforms previous hierarchical feature selection methods and also traditional methods in terms of predictive accuracy, selecting smaller feature subsets in general.

Automated Machine Learning (AutoML) has emerged to deal with the selection and configuration of algorithms for a given learning task. With the progression of AutoML, several effective methods were introduced, especially for traditional classification and regression problems. Apart from the AutoML success, several issues remain open. One issue, in particular, is the lack of ability of AutoML methods to deal with different types of data. Based on this scenario, this paper approaches AutoML for multi-label classification (MLC) problems. In MLC, each example can be simultaneously associated to several class labels, unlike the standard classification task, where an example is associated to just one class label. In this work, we provide a general comparison of five automated multi-label classification methods -- two evolutionary methods, one Bayesian optimization method, one random search and one greedy search -- on 14 datasets and three designed search spaces. Overall, we observe that the most prominent method is the one based on a canonical grammar-based genetic programming (GGP) search method, namely Auto-MEKA$_{GGP}$. Auto-MEKA$_{GGP}$ presented the best average results in our comparison and was statistically better than all the other methods in different search spaces and evaluated measures, except when compared to the greedy search method., Comment: GECCO'2020 paper: Submitted and accepted

Understanding the ageing process is a very challenging problem for biologists. To help in this task, there has been a growing use of classification methods (from machine learning) to learn models that predict whether a gene influences the process of ageing or promotes longevity. One type of predictive feature often used for learning such classification models is Protein-Protein Interaction (PPI) features. One important property of PPI features is their uncertainty, i.e., a given feature (PPI annotation) is often associated with a confidence score, which is usually ignored by conventional classification methods. Hence, we propose the Lazy Feature Selection for Uncertain Features (LFSUF) method, which is tailored for coping with the uncertainty in PPI confidence scores. In addition, following the lazy learning paradigm, LFSUF selects features for each instance to be classified, making the feature selection process more flexible. We show that our LFSUF method achieves better predictive accuracy when compared to other feature selection methods that either do not explicitly take PPI confidence scores into account or deal with uncertainty globally rather than using a per-instance approach. Also, we interpret the results of the classification process using the features selected by LFSUF, showing that the number of selected features is significantly reduced, assisting the interpretability of the results. The datasets used in the experiments and the program code of the LFSUF method are freely available on the web at http://github.com/pablonsilva/FSforUncertainFeatureSpaces.

Most surgical specialties have attempted to address concerns about the unfair comparison of outcomes by 'risk-adjusting' data to benchmark specialty-specific outcomes that are indicative of quality of care. We explore the ability to predict for positive margin status so that effective benchmarking that will account for complexity of case mix is possible. A dataset of care episodes recorded as a clinical audit of margin status after surgery for head and neck squamous cell carcinoma (n=1316) was analysed within the Waikato Environment for Knowledge Analyisis (WEKA) machine learning programme. The outcome was a classification model that can predict for positivity of tumour margins (defined as less than 1mm) using data on preoperative demographics, operations, functional status, and tumour stage. Positive resection margins of less than 1mm were common, and varied considerably between treatment units (19%-29%). Four algorithms were compared to attempt to risk-adjust for case complexity. The 'champion' model was a Naive Bayes classifier (AUROC 0.72) that suggested acceptable discrimination. Calibration was good (Hosmer-Lemershow goodness-of-fit test p=0.9). Adjusted positive margin rates are presented on a funnel plot. Subspecialty groups within oral and maxillofacial surgery are seeking metrics that will allow for meaningful comparison of the quality of care delivered by surgical units in the UK. To enable metrics to be effective, we argue that they can be modelled so that meaningful benchmarking, which takes account of variation in complexity of patient need or care, is possible.

Statistical significance analysis, based on hypothesis tests, is a common approach for comparing classifiers. However, many studies oversimplify this analysis by simply checking the condition p-value < 0.05, ignoring important concepts such as the effect size and the statistical power of the test. This problem is so worrying that the American Statistical Association has taken a strong stand on the subject, noting that although the p-value is a useful statistical measure, it has been abusively used and misinterpreted. This work highlights problems caused by the misuse of hypothesis tests and shows how the effect size and the power of the test can provide important information for better decision-making. To investigate these issues, we perform empirical studies with different classifiers and 50 datasets, using the Student’s t-test and the Wilcoxon test to compare classifiers. The results show that an isolated p-value analysis can lead to wrong conclusions and that the evaluation of the effect size and the power of the test contributes to a more principled decision-making.

Motivation One way to identify genes possibly associated with ageing is to build a classification model (from the machine learning field) capable of classifying genes as associated with multiple age-related diseases. To build this model, we use a pre-compiled list of human genes associated with age-related diseases and apply a novel Deep Neural Network (DNN) method to find associations between gene descriptors (e.g. Gene Ontology terms, protein–protein interaction data and biological pathway information) and age-related diseases. Results The novelty of our new DNN method is its modular architecture, which has the capability of combining several sources of biological data to predict which ageing-related diseases a gene is associated with (if any). Our DNN method achieves better predictive performance than standard DNN approaches, a Gradient Boosted Tree classifier (a strong baseline method) and a Logistic Regression classifier. Given the DNN model produced by our method, we use two approaches to identify human genes that are not known to be associated with age-related diseases according to our dataset. First, we investigate genes that are close to other disease-associated genes in a complex multi-dimensional feature space learned by the DNN algorithm. Second, using the class label probabilities output by our DNN approach, we identify genes with a high probability of being associated with age-related diseases according to the model. We provide evidence of these putative associations retrieved from the DNN model with literature support. Availability and implementation The source code and datasets can be found at: https://github.com/fabiofabris/Bioinfo2019. Supplementary information Supplementary data are available at Bioinformatics online.

Instance-based learning (IBL) methods predict the class label of a new instance based directly on the distance between the new unlabeled instance and each labeled instance in the training set, without constructing a classification model in the training phase. In this paper, we introduce a novel class-based feature weighting technique, in the context of instance-based distance methods, using the Ant Colony Optimization meta-heuristic. We address three different approaches of instance-based classification: k-Nearest Neighbours, distance-based Nearest Neighbours, and Gaussian Kernel Estimator. We present a multi-archive adaptation of the ACO? algorithm and apply it to the optimization of the key parameter in each IBL algorithm and of the class-based feature weights. We also propose an ensemble of classifiers approach that makes use of the archived populations of the ACO? algorithm. We empirically evaluate the performance of our proposed algorithms on 36 benchmark datasets, and compare them with conventional instance-based classification algorithms, using various parameter settings, as well as with a state-of-the-art coevolutionary algorithm for instance selection and feature weighting for Nearest Neighbours classifiers.

Hierarchical feature selection is a new research area in machine learning/data mining, which consists of performing feature selection by exploiting dependency relationships among hierarchically structured features. This paper evaluates four hierarchical feature selection methods, i.e., HIP, MR, SHSEL and GTD, used together with four types of lazy learning-based classifiers, i.e., Naïve Bayes, Tree Augmented Naïve Bayes, Bayesian Network Augmented Naïve Bayes and k-Nearest Neighbors classifiers. These four hierarchical feature selection methods are compared with each other and with a well-known “flat” feature selection method, i.e., Correlation-based Feature Selection. The adopted bioinformatics datasets consist of aging-related genes used as instances and Gene Ontology terms used as hierarchical features. The experimental results reveal that the HIP (Select Hierarchical Information Preserving Features) method performs best overall, in terms of predictive accuracy and robustness when coping with data where the instances’ classes have a substantially imbalanced distribution. This paper also reports a list of the Gene Ontology terms that were most often selected by the HIP method.

Most surgical specialties have attempted to address concerns about unfair comparison of outcomes by “risk-adjusting” data to benchmark specialty-specific outcomes that are indicative of the quality of care. We are building on previous work in head and neck surgery to address the current need for a robust validated means of risk adjustment. A dataset of care episodes, which were recorded as a clinical audit of complications after operations for squamous cell carcinoma (SCC) of the head and neck (n = 1254), was analysed with the Waikarto Environment for Knowledge Analysis (WEKA) machine learning tool. This produced 4 classification models that could predict complications using data on the preoperative demographics of the patients, operation, functional status, and tumour stage. Three of them performed acceptably: one that predicted “any complication” within 30 days (area under the receiver operating characteristic curve (AUROC) 0.72), one that predicted severe complications (Clavien-Dindo grade 3 or above) within 30 days (AUROC 0.70), and one that predicted a prolonged duration of hospital stay of more than 15 days, (AUROC 0.81). The final model, which was developed on a subgroup of patients who had free tissue transfer (n = 443), performed poorly (AUROC 0.59). Subspecialty groups within oral and maxillofacial surgery are seeking metrics that will allow a meaningful comparison of the quality of care delivered by surgical units in the UK. For these metrics to be effective they must show variation between units and be amendable to change by service personnel. Published baseline data must also be available. They should be modelled effectively so that meaningful comparison, which takes account of variations in the complexity of the patients’ needs or care, is possible.

Automated Machine Learning (Auto-ML) methods search for the best classification algorithm and its best hyper-parameter settings for each input dataset. Auto-ML methods normally maximize only predictive accuracy, ignoring the classification model’s interpretability – an important criterion in many applications. Hence, we propose a novel approach, based on Auto-ML, to investigate the trade-off between the predictive accuracy and the interpretability of classification-model representations. The experiments used the Auto-WEKA tool to investigate this trade-off. We distinguish between white box (interpretable) model representations and two other types of model representations: black box (non-interpretable) and grey box (partly interpretable). We consider as white box the models based on the following 6 interpretable knowledge representations: decision trees, If-Then classification rules, decision tables, Bayesian network classifiers, nearest neighbours and logistic regression. The experiments used 16 datasets and two runtime limits per Auto-WEKA run: 5 h and 20 h. Overall, the best white box model was more accurate than the best non-white box model in 4 of the 16 datasets in the 5-hour runs, and in 7 of the 16 datasets in the 20-hour runs. However, the predictive accuracy differences between the best white box and best non-white box models were often very small. If we accept a predictive accuracy loss of 1% in order to benefit from the interpretability of a white box model representation, we would prefer the best white box model in 8 of the 16 datasets in the 5-hour runs, and in 10 of the 16 datasets in the 20-hour runs.

With the ever-increasing number of pre-processing and classification algorithms, manually selecting the best algorithm and their best hyper-parameter settings (i.e. the best classification workflow) is a daunting task. Automated Machine Learning (Auto-ML) methods have been recently proposed to tackle this issue. Auto-ML tools aim to automatically choose the best classification workflow for a given dataset. In this work we analyse the predictive accuracy and overfit of the state-of-the-art auto-sklearn tool, which iteratively builds a classification ensemble optimised for the user’s dataset. This work has 3 contributions. First, we measure 3 types of auto-sklearn’s overfit, involving the differences of predictive accuracies measured on different data subsets: two parts of the training set (for learning and internal validation of the model) and the hold-out test set used for final evaluation. Second, we analyse the distribution of types of classification models selected by auto-sklearn across all 17 datasets. Third, we measure correlations between predictive accuracies on different data subsets and different types of overfitting. Overall, substantial degrees of overfitting were found in several datasets, and decision tree ensembles were the most frequently selected types of models.

This study analyzed the effects of photobiomodulation (PBM) with low-level laser therapy on nociceptive behavior and neuronal activity in the trigeminal nucleus after experimental unilateral temporomandibular joint (TMJ) disc injury. The animals were divided into 4 groups (n = 10 each): group 1, surgical injury of the articular disc and PBM; group 2, sham-operated subjected to PBM; group 3, surgical injury of the articular disc; and group 4, control (Naive). Ten sessions of PBM were performed using GaAs laser with a wavelength of 904 nm, power of 75 W pico, average power of 0.043 W, area of the beam of 0.13 cm2, duration of the pulses of 60 nseg (in the frequency of 9500 Hz), energy density of 5.95 J/cm2, energy per point of 0.7 J, and power density of 333.8 mW/cm2, and the irradiation was done for 18 s per point. Neuropathic symptoms were evaluated using the von Frey test. Trigeminal ganglion samples underwent immunoblotting to examine the expression of substance P, vanilloid transient potential receptor of subtype-1 (TRPV-1), and peptide related to the calcitonin gene (CGRP). There was a total decrease in pain sensitivity after the second session of PBM in operated animals, and this decrease remains until the last session. There was a significant decrease in the expression of SP, TRPV-1, and CGRP after PBM. Photobiomodulation therapy was effective in reducing nociceptive behavior and trigeminal nucleus neuronal activity after TMJ disc injury.

Automated Machine Learning (Auto-ML) is an emerging area of ML which consists of automatically selecting the best ML algorithm and its best hyper-parameter settings for a given input dataset, by doing a search in a large space of candidate algorithms and settings. In this work we propose a new Evolutionary Algorithm (EA) for the Auto-ML task of automatically selecting the best ensemble of classifiers and their hyper-parameter settings for an input dataset. The proposed EA was compared against a version of the well-known Auto-WEKA method adapted to search in the same space of algorithms and hyper-parameter settings as the EA. In general, the EA obtained significantly smaller classification error rates than that Auto-WEKA version in experiments with 15 classification datasets.

Most ant colony optimization (ACO) algorithms for inducing classification rules use a ACO-based procedure to create a rule in a one-at-a-time fashion. An improved search strategy has been proposed in the cAnt-Miner[Formula: see text] algorithm, where an ACO-based procedure is used to create a complete list of rules (ordered rules), i.e., the ACO search is guided by the quality of a list of rules instead of an individual rule. In this paper we propose an extension of the cAnt-Miner[Formula: see text] algorithm to discover a set of rules (unordered rules). The main motivations for this work are to improve the interpretation of individual rules by discovering a set of rules and to evaluate the impact on the predictive accuracy of the algorithm. We also propose a new measure to evaluate the interpretability of the discovered rules to mitigate the fact that the commonly used model size measure ignores how the rules are used to make a class prediction. Comparisons with state-of-the-art rule induction algorithms, support vector machines, and the cAnt-Miner[Formula: see text] producing ordered rules are also presented.

In recent years, multi-label classification (MLC) has become an emerging research topic in big data analytics and machine learning. In this problem, each object of a dataset may belong to multiple class labels and the goal is to learn a classification model that can infer the correct labels of new, previously unseen, objects. This paper presents a survey of genetic algorithms (GAs) designed for MLC tasks. The study is organized in three parts. First, we propose a new taxonomy focused on GAs for MLC. In the second part, we provide an up-to-date overview of the work in this area, categorizing the approaches identified in the literature with respect to the taxonomy. In the third and last part, we discuss some new ideas for combining GAs with MLC.

Motivation This work uses the Random Forest (RF) classification algorithm to predict if a gene is over-expressed, under-expressed or has no change in expression with age in the brain. RFs have high predictive power, and RF models can be interpreted using a feature (variable) importance measure. However, current feature importance measures evaluate a feature as a whole (all feature values). We show that, for a popular type of biological data (Gene Ontology-based), usually only one value of a feature is particularly important for classification and the interpretation of the RF model. Hence, we propose a new algorithm for identifying the most important and most informative feature values in an RF model. Results The new feature importance measure identified highly relevant Gene Ontology terms for the aforementioned gene classification task, producing a feature ranking that is much more informative to biologists than an alternative, state-of-the-art feature importance measure. Availability and implementation The dataset and source codes used in this paper are available as ‘Supplementary Material’ and the description of the data can be found at: https://fabiofabris.github.io/bioinfo2018/web/. Supplementary information Supplementary data are available at Bioinformatics online.

We propose a new variant of the Correlation-based\ud Feature Selection (CFS) method for coping with longitudinal data\ud where variables are repeatedly measured across different time\ud points. The proposed CFS variant is evaluated on ten datasets\ud created using data from the English Longitudinal Study of Ageing\ud (ELSA), with different age-related diseases used as the class\ud variables to be predicted. The results show that, overall, the\ud proposed CFS variant leads to better predictive performance than\ud the standard CFS and the baseline approach of no feature\ud selection, when using Naïve Bayes and J48 decision tree induction\ud as classification algorithms (although the difference in\ud performance is very small in the results for J4.8). We also report\ud the most relevant features selected by J48 across the datasets.

Objectives: Temporomandibular disorders (TMD) may present with diverse symptoms and multiple signs. Thus, the use of indices allows reliable quantification of the signs and symptoms of the TMD base...

Measurements of maturity depend on the biological system considered since differences are often found in performance and body size in subjects of the same chronological age. The objective of this study was to identify associations between biological maturation, body morphology and physical performance in girls aged from 8.0 to 15.9 year-old and to verify the bone age in obese girls and compare it with chronological age. For that purpose 2040 (11.9 ± 2.3 years) school girls from Montes Claros, participated in this study. Regular anthropometric measures as height and body mass were taken. Triceps, biceps, subscapular, abdominal, suprailiac and calf skinfolds were also registered. Physical performance was assessed trough the test of a standing long jump, handgrip strength and 20 m multistage shuttle run. Maturational status, the average age at menarche and identification of PHV (maturity off set) were determined by means of the retrospective method. Girls with the BMI above the 95th percentile got their bone age evaluated through X-ray of the left hand/wrist, in accordance with the FELS method. It was possible to find an average age at menarche of 11.30 ± 0.70, while the average age at PHV was 12.17 ± 0.71 years of age. It was observed that both body composition and physical performance showed a tendency to increase with advancing age. However, when controlling the effect of maturation, despite having higher values in body composition the post-menarche girls group did not show higher levels of physical performance. In all age groups, obese girls showed mean rates of bone age higher than chronologic age (12.25 ± 2.09 and 14.09 ± 2.35, respectively, p=0.000). Chronological age should be used with caution when evaluating obese teenagers as it may underestimate biological age.

The biopharmaceutical classification system (BCS) is now well established and utilized for the development and biowaivers of immediate oral dosage forms. The prediction of BCS class can be carried out using multilabel classification. Unlike single label classification, multilabel classification methods predict more than one class label at the same time. This paper compares two multilabel methods, binary relevance and classifier chain, for provisional BCS class prediction. Large data sets of permeability and solubility of drug and drug-like compounds were obtained from the literature and were used to build models using decision trees. The separate permeability and solubility models were validated, and a BCS validation set of 127 compounds where both permeability and solubility were known was used to compare the two aforementioned multilabel classification methods for provisional BCS class prediction. Overall, the results indicate that the classifier chain method, which takes into account label interactions, performed better compared to the binary relevance method. This work offers a comparison of multilabel methods and shows the potential of the classifier chain multilabel method for improved biological property predictions for use in drug discovery and development.

Given a new dataset for classification in Machine Learning (ML), finding the best classification algorithm and the best configuration of its (hyper)-parameters for that particular dataset is an open issue. The Automatic ML (Auto-ML) area has emerged to solve this task. With this issue in mind, in this work we are interested in a specific type of classification problem, called multi-label classification (MLC). In MLC, each example in the dataset can be associated to one or more class labels, making the task considerably harder than traditional, single-label classification. In addition, the cost of learning raises due to the higher complexity of the data. Although the literature has proposed some methods to solve the Auto-ML task, those methods address only the traditional, single-label classification problem. By contrast, this work proposes the first method (an evolutionary algorithm) for solving the Auto-ML task in MLC, i.e., the first method for automatically selecting and configuring the best MLC algorithm for a given input dataset. The proposed evolutionary algorithm is evaluated on three MLC datasets, and compared against two baseline methods according to four different multi-label predictive accuracy measures. The results show that the proposed evolutionary algorithm is competitive against the baselines, but there is still room for improvement.

Regression problems provide some of the most challenging research opportunities in the area of machine learning, and more broadly intelligent systems, where the predictions of some target variables are critical to a specific application. Rainfall is a prime example, as it exhibits unique characteristics of high volatility and chaotic patterns that do not exist in other time series data. This work’s main impact is to show the benefit machine learning algorithms, and more broadly intelligent systems have over the current state-of-the-art techniques for rainfall prediction within rainfall derivatives. We apply and compare the predictive performance of the current state-of-the-art (Markov chain extended with rainfall prediction) and six other popular machine learning algorithms, namely: Genetic Programming, Support Vector Regression, Radial Basis Neural Networks, M5 Rules, M5 Model trees, and k-Nearest Neighbours. To assist in the extensive evaluation, we run tests using the rainfall time series across data sets for 42 cities, with very diverse climatic features. This thorough examination shows that the machine learning methods are able to outperform the current state-of-the-art. Another contribution of this work is to detect correlations between different climates and predictive accuracy. Thus, these results show the positive effect that machine learning-based intelligent systems have for predicting rainfall based on predictive accuracy and with minimal correlations existing across climates.

Broadly speaking, supervised machine learning is the computational task of learning correlations between variables in annotated data (the training set), and using this information to create a predictive model capable of inferring annotations for new data, whose annotations are not known. Ageing is a complex process that affects nearly all animal species. This process can be studied at several levels of abstraction, in different organisms and with different objectives in mind. Not surprisingly, the diversity of the supervised machine learning algorithms applied to answer biological questions reflects the complexities of the underlying ageing processes being studied. Many works using supervised machine learning to study the ageing process have been recently published, so it is timely to review these works, to discuss their main findings and weaknesses.\ud \ud In summary, the main findings of the reviewed papers are: the link between specific types of DNA repair and ageing, ageing-related proteins tend to be highly connected and seem to play a central role in molecular pathways; ageing/longevity is linked with autophagy and apoptosis, nutrient receptor genes, and copper and iron ion transport. Additionally, several biomarkers of ageing were found by machine learning. Despite some interesting machine learning results, we also identified a weakness of current works on this topic: only one of the reviewed papers has corroborated the computational results of machine learning algorithms through wet-lab experiments. In conclusion, supervised machine learning has contributed to advance our knowledge and has provided novel insights on ageing, yet future work should have a greater emphasis in validating the predictions.

Rainfall derivatives are in their infancy since starting trading on the Chicago Mercantile Exchange (CME) since 2011. Being a relatively new class of financial instruments there is no generally recognised pricing framework used within the literature. In this paper, we propose a novel framework for pricing contracts using Genetic Programming (GP). Our novel framework requires generating a risk-neutral density of our rainfall predictions generated by GP supported by Markov chain Monte Carlo and Esscher transform. Moreover, instead of having a single rainfall model for all contracts, we propose having a separate rainfall model for each contract. We compare our novel framework with and without our proposed contract-specific models for pricing against the pricing performance of the two most commonly used methods, namely Markov chain extended with rainfall prediction (MCRP), and burn analysis (BA) across contracts available on the CME. Our goal is twofold, (i) to show that by improving the predictive accuracy of the rainfall process, the accuracy of pricing also increases. (ii) contract-specific models can further improve the pricing accuracy. Results show that both of the above goals are met, as GP is capable of pricing rainfall futures contracts closer to the CME than MCRP and BA. This shows that our novel framework for using GP is successful, which is a significant step forward in pricing rainfall derivatives.

Bayesian multi-net (BMN) classifiers consist of several local models, one for each data subset, to model asymmetric, more consistent dependency relationships among variables in each subset. This paper extends an earlier work of ours and proposes several contributions to the field of clustering-based BMN classifiers, using Ant Colony Optimisation (ACO). First, we introduce a new medoid-based method for ACO-based clustering in the Ant-ClustBMB algorithm to learn BMNs. Both this algorithm and our previously introduced Ant-ClustBIB for instance-based clustering have their effectiveness empirically compared in the context of the “cluster-then-learn” approach, in which the ACO clustering step completes before learning the local BN classifiers. Second, we propose a novel “cluster-with-learn” approach, in which the ACO meta-heuristic performs the clustering and the BMN learning in a synergistic fashion. Third, we adopt the latter approach in two new ACO algorithms: ACO-ClustBIB, using the instance-based method, and ACO-ClustBMB, using the medoid-based method. Empirical results are obtained on 30 UCI datasets.

The vast majority of the literature evaluates the performance of classification models using only the criterion of predictive accuracy. This paper reviews the case for considering also the comprehensibility (interpretability) of classification models, and discusses the interpretability of five types of classification models, namely decision trees, classification rules, decision tables, nearest neighbors and Bayesian network classifiers. We discuss both interpretability issues which are specific to each of those model types and more generic interpretability issues, namely the drawbacks of using model size as the only criterion to evaluate the comprehensibility of a model, and the use of monotonicity constraints to improve the comprehensibility and acceptance of classification models by users.

Bayesian networks are knowledge representation tools that model the (in)dependency relationships among variables for probabilistic reasoning. Classification with Bayesian networks aims to compute the class with the highest probability given a case. This special kind is referred to as Bayesian network classifiers. Since learning the Bayesian network structure from a dataset can be viewed as an optimization problem, heuristic search algorithms may be applied to build high-quality networks in medium- or large-scale problems, as exhaustive search is often feasible only for small problems. In this paper, we present our new algorithm, ABC-Miner, and propose several extensions to it. ABC-Miner uses ant colony optimization for learning the structure of Bayesian network classifiers. We report extended computational results comparing the performance of our algorithm with eight other classification algorithms, namely six variations of well-known Bayesian network classifiers, cAnt-Miner for discovering classification rules and a support vector machine algorithm.

Hierarchical multilabel classification is a complex classification problem where an instance can be assigned to more than one class simultaneously, and these classes are hierarchically organized with superclasses and subclasses, that is, an instance can be classified as belonging to more than one path in the hierarchical structure. This article experimentally analyses the behavior of different decision tree-based hierarchical multilabel classification methods based on the local and global classification approaches. The approaches are compared using distinct hierarchy-based and distance-based evaluation measures, when they are applied to a variation of real multilabel and hierarchical datasets' characteristics. Also, the different evaluation measures investigated are compared according to their degrees of consistency, discriminancy, and indifferency. As a result of the experimental analysis, we recommend the use of the global classification approach and suggest the use of the Hierarchical Precision and Hierarchical Recall evaluation measures.

Ant colony optimization (ACO) algorithms have been successfully applied to discover a list of classification rules. In general, these algorithms follow a sequential covering strategy, where a single rule is discovered at each iteration of the algorithm in order to build a list of rules. The sequential covering strategy has the drawback of not coping with the problem of rule interaction, i.e., the outcome of a rule affects the rules that can be discovered subsequently since the search space is modified due to the removal of examples covered by previous rules. This paper proposes a new sequential covering strategy for ACO classification algorithms to mitigate the problem of rule interaction, where the order of the rules is implicitly encoded as pheromone values and the search is guided by the quality of a candidate list of rules. Our experiments using 18 publicly available data sets show that the predictive accuracy obtained by a new ACO classification algorithm implementing the proposed sequential covering strategy is statistically significantly higher than the predictive accuracy of state-of-the-art rule induction classification algorithms.

Regression problems provide some of the most challenging research opportunities, where the predictions of such domains are critical to a specific application. Problem domains that exhibit large variability and are of chaotic nature are the most challenging to predict. Rainfall being a prime example, as it exhibits very unique characteristics that do not exist in other time series data. Moreover, rainfall is essential for applications that surround financial securities such as rainfall derivatives. This paper is interested in creating a new methodology for increasing the predictive accuracy of rainfall within the problem domain of rainfall derivatives. Currently, the process of predicting rainfall within rainfall derivatives is dominated by statistical models, namely Markov-chain extended with rainfall prediction (MCRP). In this paper, we propose a novel algorithm for decomposing rainfall, which is a hybrid Genetic Programming/Genetic Algorithm (GP/GA) algorithm. Hence, the overall problem becomes easier to solve. We compare the performance of our hybrid GP/GA, against MCRP, Radial Basis Function and GP without decomposition. We aim to show the effectiveness that a decomposition algorithm can have on the problem domain. Results show that in general decomposition has a very positive effect by statistically outperforming GP without decomposition and MCRP.

Rainfall is one of the most challenging variables to predict, as it exhibits very unique characteristics that do not exist in other time series data. Moreover, rainfall is a major component and is essential for applications that surround water resource planning. In particular, this paper is interested in extending previous work carried out on the prediction of rainfall using Genetic Programming (GP) for rainfall derivatives. Currently in the rainfall derivatives literature, the process of predicting rainfall is dominated by statistical models, namely using a Markov-chain extended with rainfall prediction (MCRP). In this paper we further extend our new methodology by looking at the effect of feature engineering on the rainfall prediction process. Feature engineering will allow us to extract additional information from the data variables created. By incorporating feature engineering techniques we look to further tailor our GP to the problem domain and we compare the performance of the previous GP, which previously statistically outperformed MCRP, against our new GP using feature engineering on 21 different data sets of cities across Europe and report the results. The goal is to see whether GP can outperform its predecessor without extra features, which acts as a benchmark. Results indicate that in general GP using extra features significantly outperforms a GP without the use of extra features.

In kwonledge extraction from databases, a frequently found problem is noisy data. Neural networks are usually tolerant a noise in the training set, but they are have a poor performance in explaining how a solution is found. In this work it is presented a method for obtainiy correct and comprehensible kwonledge by using rule extraction from trained neural networks. Such a method uses genetic algorithm to find a suitable topology for a neural network that allows the RX algorithm to extract a rule set as accurate and comprehensible as possible. The proposed system was tested with two publicly available data sets and results were very satisfactory.

Motivation: The incidence of ageing-related diseases has been constantly increasing in the last decades, raising the need for creating effective methods to analyze ageing-related protein data. These methods should have high predictive accuracy and be easily interpretable by ageing experts. To enable this, one needs interpretable classification models (supervised machine learning) and features with rich biological meaning. In this paper we propose two interpretable feature types based on Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and compare them with traditional feature types in hierarchical classification (a more challenging classification task regarding predictive performance) and binary classification (a classification task producing easier to interpret classification models). As far as we know, this work is the first to: (i) explore the potential of the KEGG pathway data in the hierarchical classification setting, (i) use the graph structure of KEGG pathways to create a feature type that quantifies the influence of a current protein on another specific protein within a KEGG pathway graph and (iii) propose a method for interpreting the classification models induced using KEGG features. Results: We performed tests measuring predictive accuracy considering hierarchical and binary class labels extracted from the Mouse Phenotype Ontology. One of the KEGG feature types leads to the highest predictive accuracy among five individual feature types across three hierarchical classification algorithms. Additionally, the combination of the two KEGG feature types proposed in this work results in one of the best predictive accuracies when using the binary class version of our datasets, at the same time enabling the extraction of knowledge from ageing-related data using quantitative influence information. Availability and Implementation: The datasets created in this paper will be freely available after publication. Contact: ff79@kent.ac.uk Supplementary information: Supplementary data are available at Bioinformatics online.

Decision trees have been widely used in data mining and machine learning as a comprehensible knowledge representation. While ant colony optimization (ACO) algorithms have been successfully applied to extract classification rules, decision tree induction with ACO algorithms remains an almost unexplored research area. In this paper we propose a novel ACO algorithm to induce decision trees, combining commonly used strategies from both traditional decision tree induction algorithms and ACO. The proposed algorithm is compared against three decision tree induction algorithms, namely C4.5, CART and cACDT, in 22 publicly available data sets. The results show that the predictive accuracy of the proposed algorithm is statistically significantly higher than the accuracy of both C4.5 and CART, which are well-known conventional algorithms for decision tree induction, and the accuracy of the ACO-based cACDT decision tree algorithm.

This paper presents a survey of evolutionary algorithms that are designed for decision-tree induction. In this context, most of the paper focuses on approaches that evolve decision trees as an alternate heuristics to the traditional top-down divide-and-conquer approach. Additionally, we present some alternative methods that make use of evolutionary algorithms to improve particular components of decision-tree classifiers. The paper's original contributions are the following. First, it provides an up-to-date overview that is fully focused on evolutionary algorithms and decision trees and does not concentrate on any specific evolutionary approach. Second, it provides a taxonomy, which addresses works that evolve decision trees and works that design decision-tree components by the use of evolutionary algorithms. Finally, a number of references are provided that describe applications of evolutionary algorithms for decision-tree induction in different domains. At the end of this paper, we address some important issues and open questions that can be the subject of future research.

This study comprehensively evaluates the performance of five types of probabilistic hierarchical classification methods used for predicting Gene Ontology (GO) terms related to ageing. Of those tested, a new hybrid of a Local Hierarchical Classifier (LHC) and the Predictive Clustering Tree algorithm (LHC-PCT) had the best predictive accuracy results. We also tested the impact of two types of variations in most hierarchical classification algorithms, namely: (a) changing the base algorithm (we tested Naive Bayes and Support Vector Machines), and the impact of (b) using or not the Correlation based Feature Selection (CFS) algorithm in a pre-processing step. In total, we evaluated the predictive performance of 17 variations of hierarchical classifiers across 15 datasets of ageing and longevity-related genes. We conclude that the LHC-PCT algorithm ranks better across several tests (seven out of 12). In addition, we interpreted the models generated by the PCT algorithm to show how hierarchical classification algorithms can be used to extract biological insights out of the ageing-related datasets that we compiled.

Rainfall is one of the most challenging variables to predict, as it exhibits very unique characteristics that do not exist in other time series data. Moreover, rainfall is a major component and is essential for applications that surround water resource planning. In particular, this paper is interested in the prediction of rainfall for rainfall derivatives. Currently in the rainfall derivatives literature, the process of predicting rainfall is dominated by statistical models, namely using a Markov-chain extended with rainfall prediction (MCRP). In this paper we outline a new methodology to be carried out by predicting rainfall with Genetic Programming (GP). This is the first time in the literature that GP is used within the context of rainfall derivatives. We have created a new tailored GP to this problem domain and we compare the performance of the GP and MCRP on 21 different data sets of cities across Europe and report the results. The goal is to see whether GP can outperform MCRP, which acts as a benchmark. Results indicate that in general GP significantly outperforms MCRP, which is the dominant approach in the literature.

Automatically inferring the function of unknown proteins is a challenging task in proteomics. There are two major problems in the task of computational protein function prediction, which are the choice of the protein representation and the choice of the classification algorithm. There are several ways of extracting features from a protein, and the choice of the feature representation might be as important as the choice of the classification algorithm. These problems are aggravated in the case of hierarchical protein function prediction, where a hierarchy of classifiers is built and each of those classifiers' construction has to consider the aforementioned selection problems. In this paper we address these problem by employing three alternative selective hierarchical classification approaches: a selecting the best classifier given a fixed representation; b selecting the best representation given a fixed classifier; and c selecting the best classifier and representation simultaneously, in a synergistic fashion. The analysis of the results have shown that the selective representation approach is almost always ranked number 1 when compared against the different fixed representations and that the use of the selective classifier approach is not able to surpass using only the best classifier for the target problem.

Attribute selection is a data preprocessing step which aims at identifying relevant attributes for the target machine learning task --namely classification in this paper. In this paper, we propose a new attribute selection strategy --based on a lazy learning approach --which postpones the identification of relevant attributes until an instance is submitted for classification. Our strategy relies on the hypothesis that taking into account the attribute values of an instance to be classified may contribute to identifying the best attributes for the correct classification of that particular instance. Experimental results using the k-NN and Naive Bayes classifiers, over 40 different data sets from the UCI Machine Learning Repository and five large data sets from the NIPS 2003 feature selection challenge, show the effectiveness of delaying attribute selection to classification time. The proposed lazy technique in most cases improves the accuracy of classification, when compared with the analogous attribute selection approach performed as a data preprocessing step. We also propose a metric to estimate when a specific data set can benefit from the lazy attribute selection approach.