Simple Classification Rules Research Articles

Generating simple classification rules to predict local surges in COVID-19 hospitalizations

Low rates of vaccination, emergence of novel variants of SARS-CoV-2, and increasing transmission relating to seasonal changes and relaxation of mitigation measures leave many US communities at risk for surges of COVID-19 that might strain hospital capacity, as in previous waves. The trajectories of COVID-19 hospitalizations differ across communities depending on their age distributions, vaccination coverage, cumulative incidence, and adoption of risk mitigating behaviors. Yet, existing predictive models of COVID-19 hospitalizations are almost exclusively focused on national- and state-level predictions. This leaves local policymakers in urgent need of tools that can provide early warnings about the possibility that COVID-19 hospitalizations may rise to levels that exceed local capacity. In this work, we develop a framework to generate simple classification rules to predict whether COVID-19 hospitalization will exceed the local hospitalization capacity within a 4- or 8-week period if no additional mitigating strategies are implemented during this time. This framework uses a simulation model of SARS-CoV-2 transmission and COVID-19 hospitalizations in the US to train classification decision trees that are robust to changes in the data-generating process and future uncertainties. These generated classification rules use real-time data related to hospital occupancy and new hospitalizations associated with COVID-19, and when available, genomic surveillance of SARS-CoV-2. We show that these classification rules present reasonable accuracy, sensitivity, and specificity (all ≥ 80%) in predicting local surges in hospitalizations under numerous simulated scenarios, which capture substantial uncertainties over the future trajectories of COVID-19. Our proposed classification rules are simple, visual, and straightforward to use in practice by local decision makers without the need to perform numerical computations.

NEW DISCRETE CROW SEARCH ALGORITHM FOR CLASS ASSOCIATION RULE MINING

Associative Classification (AC) or Class Association Rule (CAR) mining is a very efficient method for the classification problem. It can build comprehensible classification models in the form of a list of simple IF-THEN classification rules from the available data. In this paper, we present a new, and improved discrete version of the Crow Search Algorithm (CSA) called NDCSA-CAR to mine the Class Association Rules. The goal of this article is to improve the data classification accuracy and the simplicity of classifiers. The authors applied the proposed NDCSA-CAR algorithm on eleven benchmark dataset and compared its result with traditional algorithms and recent well known rule-based classification algorithms. The experimental results show that the proposed algorithm outperformed other rule-based approaches in all evaluated criteria.

New Discrete Crow Search Algorithm for Class Association Rule Mining

Associative Classification (AC) or Class Association Rule (CAR) mining is a very efficient method for the classification problem. It can build comprehensible classification models in the form of a list of simple IF-THEN classification rules from the available data. In this paper, we present a new, and improved discrete version of the Crow Search Algorithm (CSA) called NDCSA-CAR to mine the Class Association Rules. The goal of this article is to improve the data classification accuracy and the simplicity of classifiers. The authors applied the proposed NDCSA-CAR algorithm on eleven benchmark dataset and compared its result with traditional algorithms and recent well known rule-based classification algorithms. The experimental results show that the proposed algorithm outperformed other rule-based approaches in all evaluated criteria.

A Decision-Making tool based on historical data for service time prediction in outpatient scheduling

BackgroundAppointment scheduling in outpatient settings typically uses simple classification rules to assign patients to long or short appointment slots, based on the anticipated duration of the patient-physician consultation, i.e., the service time. For example, new patients are assigned longer appointment slots, and return patients are assigned shorter slots. While these rules are convenient, they fail to account for the significant variability in service time of outpatient visits. MethodsWe present a data-mining approach that allows practices to predict service time based on patient characteristics and several other clinical attributes. This approach provides a decision-support tool that helps practices determine the length of time to allocate to a patient’s appointment. Specifically, we use a neural network to accurately estimate service time for each patient based on his/her characteristics. The neural network is trained using eight years of real appointment data (2010 to 2018) from a local outpatient clinic. We compare the performance of the neural network predictions against commonly used classification rules, using a randomly sampled test dataset and a statistical test. ResultsOur results suggest that outpatient practices can refine their current practices by adopting a data-driven approach to determining slot lengths for appointments. The average absolute difference and the standard deviation of differences between the neural network predictions and the actual service times in practice (case study) are 5.7 min and 4.0 min, respectively. These two measures are significantly lower than the same comparison with the common classification rule (new patient versus return patient) at the clinic; i.e. average time and standard deviations are 14.3 min and 8.2 min, respectively. ConclusionNeural network modeling can capture the effect of processes in a medical facility and create individualized predictions of patient service time with more accuracy.

A CFAR Detection Approach for Identifying Gas Bubble Seeps With Multibeam Echo Sounders

A cell-averaged constant false alarm rate (CFAR) detector is described and applied to data collected with a multibeam echo sounder (MBES). The CFAR detector is designed specifically for transient targets observed with MBES, and operates under the assumption that background noise, including volume and seafloor reverberation, is locally stationary in time. This assumption, and the CFAR detector performance in general, was examined for data collected by a 30-kHz MBES operating in the Gulf of Mexico where the targets of interest were methane gas bubble plumes rising up from the seabed. Results with example data suggest that the CFAR detector was able to remove 99.1% of the MBES raw data while preserving the targets of interest. False detections appear randomly distributed throughout a single MBES ping, unlike the targets, and a within-ping target clustering algorithm was able to remove many of the false detections. In a single ping, an example is shown where the combined CFAR detector and a target cluster-size rule was able to reduce the number of false detections to 99.8% of the original data. The detector and cluster-size rules were applied to a sequence of approximately 400 pings, and two additional morphological rules based on the size and aspect ratio of the resulting target clusters were then applied to the detections to isolate the MBES backscatter intensity associated with gas bubbles. This combination of CFAR detector and simple morphological classification rules provides a useful way to detect gas bubble seeps or other transient targets.

Learning Multiple Parameters for Kernel Collaborative Representation Classification.

In this article, the problem of automatically learning multiple parameters for kernel collaborative representation classification (KCRC) is considered. We investigate the KCRC and measure its generalization error via leave-one-out cross-validation (LOO-CV). By taking advantage of the specific properties of KCRC, a closed-form expression is derived for the outputs of LOO-CV. Then, a simple classification rule that provides probabilistic outputs is adopted, and thereby, an effective loss function that is an explicit function with respect to the parameters is proposed as the generalization error. The gradients of the loss function are calculated, and the parameters are learned by minimizing the loss function using a gradient-based optimization algorithm. Furthermore, the proposed approach makes it possible to solve the multiple kernel/feature learning problems of KCRC effectively. Experiment results on six data sets taken from different scenes demonstrate the effectiveness of the proposed approach.

Estimation of Diabetes in a High-Risk Adult Chinese Population Using J48 Decision Tree Model.

BackgroundTo predict and make an early diagnosis of diabetes is a critical approach in a population with high risk of diabetes, one of the devastating diseases globally. Traditional and conventional blood tests are recommended for screening the suspected patients; however, applying these tests could have health side effects and expensive cost. The goal of this study was to establish a simple and reliable predictive model based on the risk factors associated with diabetes using a decision tree algorithm.MethodsA retrospective cross-sectional study was used in this study. A total of 10,436 participants who had a health check-up from January 2017 to July 2017 were recruited. With appropriate data mining approaches, 3454 participants remained in the final dataset for further analysis. Seventy percent of these participants (2420 cases) were then randomly allocated to either the training dataset for the construction of the decision tree or the testing dataset (30%, 1034 cases) for evaluation of the performance of the decision tree. For this purpose, the cost-sensitive J48 algorithm was used to develop the decision tree model.ResultsUtilizing all the key features of the dataset consisting of 14 input variables and two output variables, the constructed decision tree model identified several key factors that are closely linked to the development of diabetes and are also modifiable. Furthermore, our model achieved an accuracy of classification of 90.3% with a precision of 89.7% and a recall of 90.3%.ConclusionBy applying simple and cost-effective classification rules, our decision tree model estimates the development of diabetes in a high-risk adult Chinese population with strong potential for implementation of diabetes management.

Designing genetic programming classifiers with feature selection and feature construction

Due to the flexibility of Genetic Programming (GP), GP has been used for feature construction, feature selection and classifier construction. In this paper, GP classifiers with feature selection and feature construction are investigated to obtain simple and effective classification rules. During the construction of a GP classifier, irrelevant and redundant features affect the search ability of GP, and make GP easily fall into local optimum. This paper proposes two new GP classifier construction methods to restrict bad impact of irrelevant and redundant features on GP classifier. The first is to use a multiple-objective fitness function that decreases both classification error rate and the number of selected features, which is named as GPMO. The second is to first use a feature selection method, i.e., linear forward selection (LFS) to remove irrelevant and redundant features and then use GPMO to construct classifiers, which is named as FSGPMO. Experiments on twelve datasets show that GPMO and FSGPMO have advantages over GP classifiers with a single-objective fitness function named GPSO in term of classification performance, the number of selected features, time cost and function complexity. The proposed FSGPMO can achieve better classification performance than GPMO on higher dimension datasets, however, FSGPMO may remove potential effective features for GP classifier and achieve much lower classification performance than GPMO on some datasets. Compared with two other GP-based classifiers, GPMO can significantly improve the classification performance. Comparisons with other classification algorithms show that GPMO can achieve better or comparable classification performance on most selected datasets. Our proposed GPMO can achieve better performance than wrapper-based feature construction methods using GP on applications with insufficient instances. Further investigations show that bloat phenomena exists in the process of GP evolution and overfitting phenomena is not obvious. Moreover, the benefits of GP over other machine learning algorithms are discussed.

A New Practical Approach for Discrimination between Inrush Currents and Internal Faults in Power Transformers

The power transformers are protected by differential relays. These relays use the second harmonic blocking to distinguish inrush current from internal fault current. In recent years, the performance of the second harmonic blocking has decreased and therefore new methods are needed to distinguish inrush current from internal fault current. The aim of this study is to present a new method for discrimination between magnetizing inrush currents and internal faults in differential protection of power transformers. The proposed scheme is based on calculating the dq0 transformation of current signals in the abc phases of transformer terminals for different inrush current and internal fault signals. Creating new waveforms using mathematical calculations, the signals are identifiable by their characteristics. The accurate diagnosis is based on creating a classification pattern for the discrimination algorithm. To substantiate the preciseness of proposed methodology, different states of inrush and fault currents are simulated in PSCADTM/EMTDCTM. Consequently, calculations and the distinction process are carried-out in MATLAB environment. The discrimination procedure needs only current signal data of less than a quarter of power frequency cycle and uses very simple classification rules. Eventually, both simulation and experimental results show that the accuracy of proposed method is high. The results of this study confirm that simple and accurate methods which use patterns can be developed for distinguishing the internal faulty current of transformer from inrush and healthy current.

Establishing K-levels and prescribing transtibial prostheses using six-minute walk test and one-leg standing test on prosthesis: a retrospective audit.

Rehabilitation programs after amputation often include fitting a prosthesis, but prescriptions vary under similar circumstances. The US Medicare Functional Classification Level (K-level) is a scale for describing functional abilities of persons after lower-limb amputation (from 0 = no ability or potential to ambulate, to 4 = prosthetic demands of a child/active adult/athlete). Different outcome measures are used to assess K-level, including six-minute walk test (6MWT). We attempted to predict the assigned K-level of unilateral transtibial prosthesis users from their results of 6MWT and one-leg standing test on prosthesis (OLSTP). Outpatients who had been rehabilitated and fitted with transtibial prosthesis at the University Rehabilitation Institute in Ljubljana in 2014 were included in a retrospective audit. The data were analysed using receiver-operating-characteristics curves, linear discriminant analysis, classification trees and ordinal logistic regression. Among the 120 patients (aged 39-90, mean 67 years; 79% men), eight belonged to K1 level, 94 to K2, and 18 to K3 or K4; 61 could not stand on the prosthesis, eight stood on it for 1 s, and 51 stood on it for 2 s or more. With a simple classification rule based only on 6MWT (130 m threshold for K2/K3/K4 vs. K1, 340 m for K3/K4 vs. K1/K2), we observed sensitivity and specificity close to 90%. The more sophisticated statistical approaches yielded substantially similar and comparably accurate results. 6MWT and OLST could therefore be used as predictors for transtibial prosthesis prescription in clinical practice.

A new data analysis method based on feature linear combination.

In biological data, feature relationships are complex and diverse, they could reflect physiological and pathological changes. Defining simple and efficient classification rules based on feature relationships is helpful for discriminating different conditions and studying disease mechanism. The popular data analysis method, k top scoring pairs (k-TSP), explores the feature relationship by focusing on the difference of the relative level of two features in different groups and classifies samples based on the exploration. To define more efficient classification rules, we propose a new data analysis method based on the linear combination of k > 0 top scoring pairs (LC-k-TSP). LC-k-TSP applies support vector machine (SVM) to define the best linear relationship of each feature pair, scores feature pairs by the discriminative abilities of the corresponding linear combinations and selects k disjoint top scoring pairs to construct an ensemble classifier. Experiments on twelve public datasets showed the superiority of LC-k-TSP over k-TSP which evaluates the relationship of every two features in the same way. The experiment also illustrated that LC-k-TSP performed similarly to SVM and random forest (RF) in accuracy rate. LC-k-TSP studies the own unique linear combination for each feature pair and defines simple classification rules, it is easy to explore the biomedical explanation. Finally, we applied LC-k-TSP to analyze the hepatocellular carcinoma (HCC) metabolomics data and define the simple classification rules for discrimination of different liver diseases. It obtained accuracy rates of 89.76% and 89.13% in distinguishing between small HCC and hepatic cirrhosis (CIR) groups as well as between HCC and CIR groups, superior to 87.99% and 80.35% by k-TSP. Hence, defining classification rules based on feature relationships is an effective way to analyze biological data. LC-k-TSP which checks different feature pairs by their corresponding unique best linear relationship has the superiority over k-TSP which checks each pair by the same linear relationship. Availability and implementation: http://www.402.dicp.ac.cn/download_ok_4.htm.

A general moving detection method using dual-target nonparametric background model

Designing a general motion detection method that has self-adaptive parameters remains a challenging issue in video surveillance. To address this problem, in this paper, a dual-target nonparametric background modeling (DTNBM) method is proposed. This model integrates the gray value and gradient to represent each pixel, which enhances the discriminative ability of the background model. We design a simple but effective classification rule for determining whether a pixel belongs to a motionless object or dynamic background. Moreover, DTNBM provides suitable updating strategies for the two categories of pixels. Most importantly, DTNBM utilizes a dual-target updating strategy to preserve the completeness of static objects and prevent false detections that are caused by background initialization or frequent background variations. To improve the updating effectiveness and efficiency, we combine similar and random schemes for background updating. The key features of DTNBM include nonparametric modeling and a controlling threshold adaptation process, which render our method easy to use on various scenarios. Comprehensive experiments have been conducted, and the results demonstrate that DTNBM outperforms the state-of-the-art methods in foreground detection.

Background modelling using discriminative motion representation

Robustness is an important factor for background modelling on various scenarios. Current pixel‐based adaptive segmentation method cannot effectively tackle diverse objects simultaneously. To address this problem, in this study, a background modelling method using discriminative motion representation is proposed. Instead of simple usage of intensity to construct the background model, the proposed method extracts a new local descriptor which uses a weighted combination of differential excitations for each pixel to enhance the discriminability of pixels. On the basis of this background model, different categories of objects can be quickly identified by a simple but effective classification rule and accurately be represented in background model by a smart selection of updating strategies. Therefore, the authors’ background modelling method can generate complete representation for static objects and decrease false detection caused by dynamic background or illumination variations. Extensive experiments have been conducted to demonstrate that the proposed method obtains more advantages of foreground detection than the state‐of‐the‐art methods. In addition, the proposed method provides a computational efficient algorithm for foreground detection tasks.

Synergy effects between grafting and subdivision in Re-RX with J48graft for the diagnosis of thyroid disease

Numerous methods for diagnosing thyroid disease have been developed, but the majority of these are black-box models. By contrast, the Recursive-Rule Extraction algorithm with J48graft is a white-box model that can provide highly accurate and concise classification rules. However, the potential capabilities of Re-RX with J48graft in terms of rule extraction remain unknown. Therefore, the aim of the present study was to elucidate the synergy effects between grafting and subdivision in Re-RX with J48graft, which work effectively in combination to extract highly accurate and concise classification rules for the diagnosis of thyroid disease. In the present study, I demonstrate how grafting and subdivision can extract highly accurate and concise classification rules from the Thyroid dataset, which is a large and highly imbalanced dataset consisting of 7200 medical records classified as normally functioning thyroid, hypothyroidism, or hyperthyroidism. I also provide the theoretical explanation underlying the excellent synergy effects between the two processes. Re-RX with J48graft not only achieved the most accurate classification rules, but also extracted simple and concrete concise classification rules for majority class samples. In addition, compared with previous methods, Re-RX with J48graft extracted rules with fewer antecedents. The maximum accuracy of the extracted rules was very high, at 97.02%. These findings suggest that Re-RX with J48graft can extract highly accurate and concise rules, which could assist healthcare professionals in the diagnosis of thyroid disease and help improve the level of care.

Block models and personalized PageRank

Methods for ranking the importance of nodes in a network have a rich history in machine learning and across domains that analyze structured data. Recent work has evaluated these methods through the "seed set expansion problem": given a subset [Formula: see text] of nodes from a community of interest in an underlying graph, can we reliably identify the rest of the community? We start from the observation that the most widely used techniques for this problem, personalized PageRank and heat kernel methods, operate in the space of "landing probabilities" of a random walk rooted at the seed set, ranking nodes according to weighted sums of landing probabilities of different length walks. Both schemes, however, lack an a priori relationship to the seed set objective. In this work, we develop a principled framework for evaluating ranking methods by studying seed set expansion applied to the stochastic block model. We derive the optimal gradient for separating the landing probabilities of two classes in a stochastic block model and find, surprisingly, that under reasonable assumptions the gradient is asymptotically equivalent to personalized PageRank for a specific choice of the PageRank parameter [Formula: see text] that depends on the block model parameters. This connection provides a formal motivation for the success of personalized PageRank in seed set expansion and node ranking generally. We use this connection to propose more advanced techniques incorporating higher moments of landing probabilities; our advanced methods exhibit greatly improved performance, despite being simple linear classification rules, and are even competitive with belief propagation.

Electronic descriptors for the antimalarial activity of sulfonamides

As an interesting class of materials for designing new antimalarial drugs, sulfonamides have shown potential for several pharmacological applications. In this study, multivariate data analyses were employed to correlate the antimalarial activity reported for a group of sulfonamide derivatives with electronic structure descriptors obtained through quantum mechanical calculations. A simple classification rule based on a discriminant function was obtained, which is able to correctly classify 94 % of the compounds as active or non-active. The obtained function combines only two electronic descriptors and provides valuable insights into the design of new derivatives with improved antimalarial potency, as well as identifies possible active sites on the structure of sulfonamides.

Application of bio-inspired krill herd algorithm for breast cancer classification and diagnosis

Our work presents a data mining approach, using Krill herd optimization algorithm to generate more comprehensive classification of breast cancer dataset. Hybrid krill herd algorithm (HKH) is proposed which is suitable for generation of optimized classification rules. HKH algorithm has a very flexible encoding where each member of population (krill herd) corresponds to a classification rule. Each krill herd consist of many krills corresponding to antecedents of classification rule. For breast cancer classification, Wisconsin diagnostic breast cancer data set created from fine needle aspiration biopsy of breast mass was used. Using this optimization algorithm a very comprehensive and simple classification rule is obtained for breast cancer classification represented in form of If-Then rule. The obtained rule can be used to further classify breast tissue specimens into malignant and benign classes, thus supporting the cancer diagnosis.

Allocating service parts in two-echelon networks at a utility company

We study a multi-item, two-echelon, continuous-review inventory problem at a Dutch utility company. We develop a model for the optimal allocation of service parts in a two-echelon network under an aggregate waiting time constraint. Specific model aspects are emergency shipments in case of stockout, and batching for regular replenishment orders at the central warehouse. We use column generation to solve this problem with various building blocks for single-item models as columns. Further, we derive simple classification rules from the solution of our multi-item, two-echelon service part optimization problem using statistical techniques. Application of our models at Liander yields a cost reduction of 15% and a decrease in the impact of waiting time for service parts by 52%.

Prediction of death in less than 60 minutes after withdrawal of cardiorespiratory support in potential organ donors after circulatory death.

Given the stable number of potential organ donors after brain death, donors after circulatory death have been an increasing source of organs procured for transplant. Among the most important considerations for donation after circulatory death (DCD) is the prediction that death will occur within a reasonable period of time after the withdrawal of cardiorespiratory support (WCRS). Accurate prediction of time to death is necessary for the procurement process. We aimed to develop simple predictive rules for death in less than 60 min and test the accuracy of these rules in a pool of potential DCD donors. A multicenter prospective longitudinal cohort design of DCD eligible patients (n=318), with the primary binary outcome being death in less than 60 min after withdrawal of cardiorespiratory support conducted in 28 accredited intensive care units (ICUs) in Australia. We used a random split-half method to produce two samples, first to develop the predictive classification rules and then to estimate accuracy in an independent sample. The best classification model used only three simple classification rules to produce an overall efficiency of 0.79 (0.72-0.85), sensitivity of 0.82 (0.73-0.90), and a positive predictive value of 0.80 (0.70-0.87) in the independent sample. Using only intensive care unit specialist prediction (a single classification rule) produced comparable efficiency 0.80 (0.73-0.86), sensitivity 0.87 (0.78-0.93), and positive predictive value 0.78 (0.68-0.86). This best predictive model missed only 18% of all potential donors. A positive prediction would be incorrect on only 20% of occasions, meaning there is an acceptable level of lost opportunity costs involved in the unnecessary assembly of transplantation teams and theatres.

Protein fold recognition based on functional domain composition

Recognition of protein fold types is an important step in protein structure and function predictions and is also an important method in protein sequence-structure research. Protein fold type reflects the topological pattern of the structure's core. Now there are three methods of protein structure prediction, comparative modeling, fold recognition and de novo prediction. Since comparative modeling is limited by sequence similarity and there is too much workload in de novo prediction, fold recognition has the greatest potential. In order to improve recognition accuracy, a recognition method based on functional domain composition is proposed in this paper. This article focuses on the 124 fold types which have more than 2 samples in LIFCA database. We apply the functional domain composition to predict the fold types of a protein or a domain. In order to evaluate our method and its sensibility to the samples involving SCOP family divided, we tested our results from different aspects. The average sensitivity, specificity and Matthew's correlation coefficient (MCC) of the 124 fold types were found to be 94.58%, 99.96% and 0.91, respectively. Our results indicate that the functional domain composition method is a very promising method for protein fold recognition. And though based on simple classification rules, LIFCA database can grasp the functional features of different proteins, reflecting the corresponding relation between protein structure and function.