Prediction of Skin Diseases Using Machine Learning

Random Forest algorithm eliminates the confines of a decision tree algorithm by reducing the over-fitting of datasets. It can handle the missing values in an effective manner and is more accurate than the decision trees. This research study presents a comparison of five different supervised machine learning algorithms namely Naive Bayes, Decision Tree, Multilayer Perceptron, Support Vector Machine and Random Forest for the classification of the disease dataset collected from an online repository. The Random Forest algorithm emerges to be the most efficient algorithm in classifying the diseases. Also, the research study applies different attribute selection measures to evaluate the impact of accuracy on the Random Forest classifier empirically. Among the different node splitting criteria, we conclude that the GINI index performed better on average than the Information Gain and Gain Ratio. The work is implemented using the Python 3.7 programming language.

Landslide susceptibility analyses using Random Forest, C4.5, and C5.0 with balanced and unbalanced datasets

Adolescents in higher education are more prone to Avoidant Personality Disorder (AVPD), which strongly affects academic achievement. The goal of this study was to create models for accurate prediction of the likelihood of Avoidant Personality Disorder among students in higher education. Information Gain, Gain Ratio, and Wrapper Approach are used as feature selection methods combined with data resampling techniques and machine learning, including Multi-Layer Perceptron Neural Network, Naïve Bayes, Decision Tree, Random Forest, and Support Vector Machine. The findings revealed that the Wrapper approach gave higher accuracy than Information Gain and Gain Ratio approach. Further, using the Gain Ratio approach gives the model a slightly higher efficiency than the Information Gain. Furthermore, when comparing feature selection and data resampling, it was found that the model using feature selection had more higher model efficiency than data resampling alone. Additionally, combining the Synthetic Minority Over-sampling Technique with Edited Nearest Neighbor (SMOTE-ENN) considerably increased the model’s effectiveness. Finally, the model’s efficiency was at its maximum, with an accuracy of 95.52%, when the Wrapper approach was used in conjunction with the Synthetic Minority Over-sampling Technique, the Edited Nearest Neighbor algorithm, and the Multi-Layer Perceptron Neural Network.

Feature transformation is an essential part of data preprocessing to improve the predictive performance of machine learning (ML) algorithms. Box-Cox transformation with the goal of separability is proven to align with the performance improvement of ML algorithms. However, the features mapped using Box-Cox transformation preserve the order of the data, so it is ineffective when used to improve the separability of multimodal distributed features. This research aims to build a feature transformation method using quadratic functions to improve class separability that can adaptively change the order of the data when necessary. Fisher score (Fs) measures the separability level by maximizing the Fisher's Criteria of the quadratic function. In addition to increasing the Fs value of each feature, this method can also make the feature more informative, as evidenced by the increasing value of information gain, information gain ratio, Gini decrease, ANOVA, Chi-Square, reliefF, and FCBF. The increase in Fs is particularly significant for bimodally distributed features. Experiments were conducted on 11 public datasets with two statistical-based machine learning algorithms representing linear and nonlinear ML algorithms to validate the success of this method, namely LDA and QDA. The experimental results show an improvement in accuracy in almost all datasets and ML algorithms, where the highest accuracy improvement is 0.268 for LDA and 0.188 for QDA.

Software industries have enormous demand for fault prediction of the faulty module and fault removal techniques. Many researchers have developed different fault prediction models to predict the fault at an early stage of the software development life cycle (SDLC). But the state-of-the-art model still suffers from the performance and generalize validation of the models. However, some researchers refer to data mining techniques, machine learning, and artificial intelligence play crucial roles in developing fault prediction models. A recent study stated that metric selection techniques also help to enhance the performance of models. Hence, to resolve the issue of improving the fault prediction model’s performance and validation, we have used data mining, instance selection, metric selection, and ensemble methods to beat the state-of-the-art results. For the validation, we have collected the 22 software projects from the four different software repositories. We have implemented three machine learning algorithms and three ensemble methods with two metric selection methods on 22 datasets. The statistical evaluation of the implemented model performed using Wilcoxon signed-rank test and the Friedman test followed by the Nemenyi test to find the significant model. As a result, the Random forest algorithm produces the best result with an average median of 95.43% (accuracy) and 0.96 (f-measure) on 22 software projects. Based on the Nemenyi test, Random forest (RF) is performing better with 4.54 (accuracy mean score) and 4.41 (f-measure mean score) shown in the critical diagram. Experimental study shows that data mining techniques with PCA provide better accuracy and f-measure.

Modeling time series of vegetation indices in tallgrass prairie using machine and deep learning algorithms

Objectives: In Japan, sodium-glucose co-transporter type 2 (SGLT2) inhibitors have been reported to be associated with serious skin and subcutaneous tissue disorders. A post-marketing surveillance (PMS) study suggested that the association was specific for ipragliflozin and, to a lesser extent for dapagliflozin. These studies were performed to confirm the association of 6 SGLT2 inhibitors with serious skin disorders in a clinical setting, to elucidate the role of melanin in serious skin disorders and to understand the underlying mechanisms.Methods: The latest PMS records were retrieved from the Japanese Adverse Drug Event Report (JADER) database, and the associations were analyzed by data mining techniques. In silico 3-D docking simulation of SGLT2 inhibitors with melanin was performed using the MOE software. The skin tissue distribution of SGLT2 inhibitors was evaluated using albino rats after oral administration at clinical doses.Results: The adjusted reporting odds ratio (95% confidential limit) was 1.667 (1.415, 1.963) for ipragliflozin, 0.514 (0.317, 0.835) for dapagliflozin, 0.149 (0.048, 0.465) for tofogliflozin, 0.624 (0.331, 1.177) for luseogliflozin, 0.590 (0.277, 1.257) for canagliflozin and 0.293 (0.073, 1.187) for empagliflozin, when drugs other than the SGLT2 inhibitors were referred, and the association was detected only for ipragliflozin in clinical use. In silico 3-D docking simulation suggested the influence of melanin in ipragliflozin-specific serious skin disorders. The skin tissue-to-plasma concentration ratio of ipragliflozin was 0.45 ± 0.20 (±SD) at 1 hr after administration and increased in a time-dependent manner to 5.82 ± 3.66 at 24 hr (p<0.05), but not in case of other SGLT2 inhibitors.Conclusions: Serious skin disorders were suggested to be specific for ipragliflozin. Interaction with melanin might be implicated in ipragliflozin-specific serious skin disorders. Ipragliflozin was retained in the skin tissue, which suggested its interaction with the skin tissue in serious skin disorders.

BackgroundDespite the interest in machine learning (ML) algorithms for analyzing real-world data (RWD) in healthcare, the use of ML in predicting time-to-event data, a common scenario in clinical practice, is less explored. ML models are capable of algorithmically learning from large, complex datasets and can offer advantages in predicting time-to-event data. We reviewed the recent applications of ML for survival analysis using RWD in healthcare.MethodsPUBMED and EMBASE were searched from database inception through March 2023 to identify peer-reviewed English-language studies of ML models for predicting time-to-event outcomes using the RWD. Two reviewers extracted information on the data source, patient population, survival outcome, ML algorithms, and the Area Under the Curve (AUC).ResultsOf 257 citations, 28 publications were included. Random survival forests (N = 16, 57%) and neural networks (N = 11, 39%) were the most popular ML algorithms. There was variability across AUC for these ML models (median 0.789, range 0.6–0.950). ML algorithms were predominately considered for predicting overall survival in oncology (N = 12, 43%). ML survival models were often used to predict disease prognosis or clinical events (N = 27, 96%) in the oncology, while less were used for treatment outcomes (N = 1, 4%).ConclusionsThe ML algorithms, random survival forests and neural networks, are mainly used for RWD to predict survival outcomes such as disease prognosis or clinical events in the oncology. This review shows that more opportunities remain to apply these ML algorithms to inform treatment decision-making in clinical practice. More methodological work is also needed to ensure the utility and applicability of ML models in survival outcomes.

PurposeIn this study, machine learning (ML) algorithms were employed to predict the shear capacity and behavior of DCSWs.Design/methodology/approachIn this study, ML algorithms were employed to predict the shear capacity and behavior of DCSWs. Various ML techniques, including linear regression (LR), support vector machine (SVM), decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost) and artificial neural network (ANN), were utilized. The ML models were trained using a dataset of 462 numerical and experimental samples. Numerical models were generated and analyzed using the finite element (FE) software Abaqus. These models underwent push-over analysis, subjecting them to pure shear conditions by applying a target displacement solely to the top of the shear walls without interaction from a frame. The input data encompassed eight survey variables: geometric values and material types. The characterization of input FE data was randomly generated within a logical range for each variable. The training and testing phases employed 90 and 10% of the data, respectively. The trained models predicted two output targets: the shear capacity of DCSWs and the likelihood of buckling. Accurate predictions in these areas contribute to the efficient lateral enhancement of structures. An ensemble method was employed to enhance capacity prediction accuracy, incorporating select algorithms.FindingsThe proposed model achieved a remarkable 98% R-score for estimating shear strength and a corresponding 98% accuracy in predicting buckling occurrences. Among all the algorithms tested, XGBoost demonstrated the best performance.Originality/valueIn this study, for the first time, ML algorithms were employed to predict the shear capacity and behavior of DCSWs.

Fluorophores play crucial roles in chemical and biological imaging. An efficient computational model that evaluates the electronic properties of molecules accurately would be a useful tool for discovering novel fluorophores. Tree‐based ensemble and graph neural network (GNN) methods have been regarded as attractive models for predicting molecular properties. Here, we present a benchmark test using three tree‐based ensemble methods (Random Forest, LightGBM, and XGBoost) and three GNNs (directed message passing neural network [D‐MPNN], attention message passing neural network [AMPNN], and DimeNet++) for predicting electronic transition properties such as excitation energies and oscillator strengths. From our benchmark, DimeNet++ was identified as the most accurate model to predict electronic transition properties. The average root mean square error of DimeNet++ for predicting the HOMO–LUMO gap was 0.11 eV whereas those of the other methods exceeded 0.3 eV. D‐MPNN predicted fastest without sacrificing accuracy. Our results show that DimeNet++ and D‐MPNN may serve as helpful evaluators for novel fluorophore design when combined with molecular generation methods.

The heart is a vital organ that is indispensable in ensuring the general health and welfare of individuals. Cardiovascular diseases (CVD) are the major health concern worldwide and a leading cause of death, leaving behind diabetes and cancer. To deal with the problem, it is essential for early detection and prediction of CVDs, which can significantly reduce morbidity and mortality rates. Computer-aided techniques facilitate physicians in the diagnosis of many heart disorders, such as valve dysfunction, heart failure, etc. Living in an "information age," every day million bytes of data are generated, and we can turn these data into knowledge for clinical investigation using the technique of data mining. Machine learning algorithms have shown promising results in predicting heart disease based on different risk parameter. In this study, for the purpose of predicting CVDs, our aim is to appraise and examine the outputs generated by machine learning algorithms including support vector machines, artificial neural network, logistic regression, random forest and decision trees.This literature survey highlights the correctness of different machine learning algorithms in forecasting heart problem and can be used as a basis for building a Clinical decision-making aid to detect and prevent heart disease at an early stage.

In recent years, the use of data mining techniques has gained significance in healthcare applications. The appropriate data mining techniques extract interesting affinities/associations between patient’s signs and symptoms, thus providing reasonable decision-making for the diagnosis and prognosis of the disease. Associative Classification (AC) is a contemporary technique, which uses the Class Association Rules (CARs) to build the classification system. Classification based on Predictive Association Rule (CPAR) is one of the popular AC algorithms, which utilise FOIL's Information Gain measure to select the best attributes for the generation of CARs. This work attempts to explore a suitable attribute selection measure and an error estimate measure that can best fit into the existing CPAR algorithm to construct an efficient rule-based classifier. Thus the performance of CPAR has been analyzed by applying alternative attribute selection measures such as Gain Ratio (GR) and Mutual Information Gain (IG) instead of FOIL’s Information Gain. Moreover, two error estimate measures such as Laplace accuracy (“La”) and Likelihood ratio statistic (“Lr”) are used for rule evaluation and best k-rule selection tasks of CPAR. This work analyzes the performance of CPAR-GR and CPAR-IG with the existing CPAR algorithm in terms of classifier accuracy. From the results, it was found that the use of “GR” and “IG” within CPAR yields accuracy higher than the existing CPAR. Significant differences in the performances of CPAR, CPAR-GR, and CPAR-IG are identified and demonstrated by experiments, and statistical tests using healthcare datasets taken from the UCI machine learning repository.

BackgroundBrain aging (BA) involves the gradual deterioration of brain systems and is associated with chronological age (CA). Measures of BA have been validated and adopted in aging and neurological disease research (Biondo, 2022; Eickhoff, 2021) and could be a useful clinical tool. BA predicts CA in healthy adults (Cole, 2017) and accelerated BA precede Alzheimer’s disease (AD) symptoms (Elliott, 2021). Current BA estimations rely on the acquisition of high‐quality neuroimaging data and complex, computationally expensive analysis workflows, creating obstacles for clinical translation. This study aims to develop an easily implementable workflow for BA where inputs are clinically relevant features derived from heterogeneously acquired magnetic resonance imaging (MRI) using the SynthSeg tool.MethodThirty‐two regional volumes from T1‐weighted MRIs from 56,505 participants (mean age=63.6; 44‐82; SD=7.52) were segmented using the SynthSeg tool. This tool analyzes clinically acquired, heterogenous inputs and requires minimal processing power, memory, and time. Only z‐transformed SynthSeg‐derived measures were included to allow investigations into confounds. Performance evaluation included three filter‐based feature selection approaches (Fischer’s Score, Information Gain, and Correlation Coefficient) combined with seven machine learning (ML) algorithms including linear regression (LR), 2nd‐order polynomial regression (2PR), Gaussian process regression (GPR), Decision Tree (DT), Random Forest (RF), a Convolutional Neural Network (CNN) treating age as a continuous variable, and as a categorical measure. Ability to predict CA was assessed through 10‐fold cross validation, each with an independent 90‐10 training‐testing split, and regression analysis between predicted and CA using an alpha=0.001.ResultFeature selection demonstrated that each structural feature contained unique information for predicting CA. All seven ML algorithms significantly predicted CA (Table 1). 2PR performed the best overall with LR and GPR following closely behind (Figure 1). Both the categorical and numerical CNN performed slightly worse, but still significantly better than the RF and DTConclusionThe association between CA and volumetric‐BA estimates derived from SynthSeg segmented data supports the potential validity of this approach. However, its simplicity may increase translatability. Future directions of this work are to improve performance. The long‐term goal of this work is to develop and validate local volumetric‐BA (as opposed to global volumetric‐BA) metrics.

In this study, our aim is to predict sex using machine learning (ML) algorithms with morphometric variables taken from the vidian canal. The study was performed on Computed Tomography (CT) images of 137 women and 137 men aged 18–65 years. Seventeen morphometric variables were determined and recorded from axial and coronal skull images. These numerical data were used for sex determination with ML algorithms. As a result of the ML algorithm analysis of the data as 80% training set and 20% test set, the highest Accuracy (Acc.) rate was 0.89 with the Extra Tree Classifier (ETC) algorithm, and the lowest was 0.78 with Quadratic Discriminant Analysis (QDA). Among the parameters, the distance between the optic canal and the vidian canal was found to have the highest effect on sex determination with the Shapley Additive Explanations analyser. As a result of 10-fold cross-validation analysis of ML algorithms, the highest Acc rate was 0.82 with Random Forest (RF) algorithm, and the lowest was 0.73 with QDA. The findings of the study reveal that morphometric variables from the vidian canal show high accuracy in sex determination.

An artificial intelligence-based decision support system for early and accurate diagnosis of Parkinson’s Disease