The Impact of Oversampling with SMOTE on the Performance of 3 Classifiers in Prediction of Type 2 Diabetes

Ramezankhani, Azra; Pournik, Omid; Shahrabi, Jamal; Azizi, Fereidoun; Hadaegh, Farzad; Khalili, Davood

doi:10.1177/0272989x14560647

Cited by 68 publications

(55 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The second category deals with disease prediction and diagnosis [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76]. Numerous algorithms and different approaches have been applied, such as traditional machine learning algorithms, ensemble learning approaches and association rule learning in order to achieve the best classification accuracy.…”

Section: Dm Through Machine Learning and Data Miningmentioning

confidence: 99%

Machine Learning and Data Mining Methods in Diabetes Research

Kavakiotis

Tsave

Salifoglou

et al. 2017

Computational and Structural Biotechnology Journal

1,059

530

View full text Add to dashboard Cite

The remarkable advances in biotechnology and health sciences have led to a significant production of data, such as high throughput genetic data and clinical information, generated from large Electronic Health Records (EHRs). To this end, application of machine learning and data mining methods in biosciences is presently, more than ever before, vital and indispensable in efforts to transform intelligently all available information into valuable knowledge. Diabetes mellitus (DM) is defined as a group of metabolic disorders exerting significant pressure on human health worldwide. Extensive research in all aspects of diabetes (diagnosis, etiopathophysiology, therapy, etc.) has led to the generation of huge amounts of data. The aim of the present study is to conduct a systematic review of the applications of machine learning, data mining techniques and tools in the field of diabetes research with respect to a) Prediction and Diagnosis, b) Diabetic Complications, c) Genetic Background and Environment, and e) Health Care and Management with the first category appearing to be the most popular. A wide range of machine learning algorithms were employed. In general, 85% of those used were characterized by supervised learning approaches and 15% by unsupervised ones, and more specifically, association rules. Support vector machines (SVM) arise as the most successful and widely used algorithm. Concerning the type of data, clinical datasets were mainly used. The title applications in the selected articles project the usefulness of extracting valuable knowledge leading to new hypotheses targeting deeper understanding and further investigation in DM.

show abstract

Section: Dm Through Machine Learning and Data Miningmentioning

confidence: 99%

Machine Learning and Data Mining Methods in Diabetes Research

Kavakiotis

Tsave

Salifoglou

et al. 2017

Computational and Structural Biotechnology Journal

1,059

530

View full text Add to dashboard Cite

show abstract

“…Assim, para trabalhos futuros pretende-se testar outras técnicas voltadas para dados desbalanceados, tanto no pré-processamento como na codificação do algoritmo classificador (ou associação de mais de um classificador), com o intuito de encontrar adaptações capazes de evitar viés para as classes majoritárias. Dentre as técnicas citadas na literatura utilizada como referência, conforme Batista, Prati e Monard (2004, 2008; e ainda Ramezankhani et al (2016) …”

Section: Resultsunclassified

Classificadores Binários, Políticas Públicas Sociais e Dados Desbalanceados

Santos¹,

Souza²,

Mayrink³

et al. 2017

RIA

View full text Add to dashboard Cite

ResumoNeste estudo utilizamos uma base de dados de pesquisa vinculada ao desempenho do Programa Bolsa Família (PBF) no ano de 2009. Este programa implica na transferência direta de renda com condicionantes nas áreas de educação, saúde e assistência social, visando atender famílias pobres e extremamente pobres -assim classificadas segundo um determinado valor percapita mensal. Esta base contém informações de cunho financeiro (renda e gastos das famílias), e também grau de instrução dos indivíduos, e elementos descritores do ambiente domiciliar (moradia e entorno). A aplicação dos algoritmos de predição visou averiguar a eficiência desses processos a partir das variáveis que descrevem as famílias, identificando corretamente se estas atendiam ou não ao perfil de beneficiárias do programa. Os algoritmos utilizados foram regressão logística, árvore binária de decisão e rede neural artificial em múltiplas camadas. Diversas medidas de desempenho foram calculadas, a partir da matriz de confusão resultante de cada algoritmo. Os valores encontrados para estas medidas foram baixos frente a uma das classes a serem identificadas. As intervenções aplicadas foram o reembaralhamento aleatório e também superamostragem da classe minoritária e sub-amostragem da classe majoritária. Embora tenha ocorrido alguma melhora, o desempenho no reconhecimento da classe minoritária permaneceu baixo o que aponta para a necessidade de novos experimentos.Palavras-chave: Árvores de decisão. Programa Bolsa Família. Classificador. Predição. Regressão logística. Redes neurais. AbstractIn this study we used a research database focused on the performance of the Bolsa Família Program (PBF) in 2009. In this program, direct income transfer is carried out under conditions of education, health and social assistance, aiming to serve poor and extremely poor families -Thus classified according to a certain monthly per capita value. This database contains financial information (income and family expenses), as well as the degree of education of the individuals, and elements describing the home environment (housing and environment). Applying prediction algorithms, the objective was to ascertain the efficiency of these processes from the variables that describe the families, correctly identifying if they met the beneficiaries profile of the program or not. The algorithms used were logistic regression, binary decision tree and artificial neural network in multiple layers. Several performance measures were calculated from the confusion matrix resulting from each algorithm. The values found for these measures were low compared to one of the classes to be identified. The interventions applied were random remarshaling and also super-sampling of the minority class and sub-sampling of the majority class. Although some improvement occurred, the performance in the minority class recognition remained low, which indicates the need for new experiments.

show abstract

“…There are a finite number of solutions to handle imbalanced data sets 21–23. In our previous work, we showed the effectiveness of Synthetic Minority Oversampling Technique (SMOTE) for handling imbalanced data sets 24. In this study, we balanced two training data sets of men and women using SMOTE as previously reported (figure 2).…”

Section: Methodsmentioning

confidence: 93%

“…In this study, we balanced two training data sets of men and women using SMOTE as previously reported (figure 2). 24…”

Section: Methodsmentioning

confidence: 99%

Decision tree-based modelling for identification of potential interactions between type 2 diabetes risk factors: a decade follow-up in a Middle East prospective cohort study

et al. 2016

Self Cite

View full text Add to dashboard Cite

ObjectiveThe current study was undertaken for use of the decision tree (DT) method for development of different prediction models for incidence of type 2 diabetes (T2D) and for exploring interactions between predictor variables in those models.DesignProspective cohort study.SettingTehran Lipid and Glucose Study (TLGS).MethodsA total of 6647 participants (43.4% men) aged >20 years, without T2D at baselines ((1999–2001) and (2002–2005)), were followed until 2012. 2 series of models (with and without 2-hour postchallenge plasma glucose (2h-PCPG)) were developed using 3 types of DT algorithms. The performances of the models were assessed using sensitivity, specificity, area under the ROC curve (AUC), geometric mean (G-Mean) and F-Measure.Primary outcome measureT2D was primary outcome which defined if fasting plasma glucose (FPG) was ≥7 mmol/L or if the 2h-PCPG was ≥11.1 mmol/L or if the participant was taking antidiabetic medication.ResultsDuring a median follow-up of 9.5 years, 729 new cases of T2D were identified. The Quick Unbiased Efficient Statistical Tree (QUEST) algorithm had the highest sensitivity and G-Mean among all the models for men and women. The models that included 2h-PCPG had sensitivity and G-Mean of (78% and 0.75%) and (78% and 0.78%) for men and women, respectively. Both models achieved good discrimination power with AUC above 0.78. FPG, 2h-PCPG, waist-to-height ratio (WHtR) and mean arterial blood pressure (MAP) were the most important factors to incidence of T2D in both genders. Among men, those with an FPG≤4.9 mmol/L and 2h-PCPG≤7.7 mmol/L had the lowest risk, and those with an FPG>5.3 mmol/L and 2h-PCPG>4.4 mmol/L had the highest risk for T2D incidence. In women, those with an FPG≤5.2 mmol/L and WHtR≤0.55 had the lowest risk, and those with an FPG>5.2 mmol/L and WHtR>0.56 had the highest risk for T2D incidence.ConclusionsOur study emphasises the utility of DT for exploring interactions between predictor variables.

show abstract

The Impact of Oversampling with SMOTE on the Performance of 3 Classifiers in Prediction of Type 2 Diabetes

Abstract: To determine a classifier with a machine learning algorithm like the PNN and DT, class skew in data should be considered. The NB and DT were optimal classifiers in a prediction task in an imbalanced medical database.

Cited by 68 publications

References 27 publications

Machine Learning and Data Mining Methods in Diabetes Research

Machine Learning and Data Mining Methods in Diabetes Research

Classificadores Binários, Políticas Públicas Sociais e Dados Desbalanceados

Decision tree-based modelling for identification of potential interactions between type 2 diabetes risk factors: a decade follow-up in a Middle East prospective cohort study

Contact Info

Product

Resources

About