Diabetes classification application with efficient missing and outliers data handling algorithms

Torkey, Hanaa; Ibrahim, Elhossiny; Hemdan, Ezz El‐Din; El‐Sayed, Ayman; Shouman, Marwa A.

doi:10.1007/s40747-021-00349-2

Cited by 8 publications

(4 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The second category includes interpretable models characterized by explicit prediction models. Most of these models rely on decision trees [ 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 , 68 , 69 ]. Although the methods based on these trees [ 70 ] could provide explicit knowledge, in many cases, it is challenging to linearize the resulting acyclic decision graphs into simple decision rules.…”

Section: State Of the Artmentioning

confidence: 99%

A Federated Learning-Inspired Evolutionary Algorithm: Application to Glucose Prediction

Falco

Cioppa

Koutný

et al. 2023

Sensors

View full text Add to dashboard Cite

In this paper, we propose an innovative Federated Learning-inspired evolutionary framework. Its main novelty is that this is the first time that an Evolutionary Algorithm is employed on its own to directly perform Federated Learning activity. A further novelty resides in the fact that, differently from the other Federated Learning frameworks in the literature, ours can efficiently deal at the same time with two relevant issues in Machine Learning, i.e., data privacy and interpretability of the solutions. Our framework consists of a master/slave approach in which each slave contains local data, protecting sensible private data, and exploits an evolutionary algorithm to generate prediction models. The master shares through the slaves the locally learned models that emerge on each slave. Sharing these local models results in global models. Being that data privacy and interpretability are very significant in the medical domain, the algorithm is tested to forecast future glucose values for diabetic patients by exploiting a Grammatical Evolution algorithm. The effectiveness of this knowledge-sharing process is assessed experimentally by comparing the proposed framework with another where no exchange of local models occurs. The results show that the performance of the proposed approach is better and demonstrate the validity of its sharing process for the emergence of local models for personal diabetes management, usable as efficient global models. When further subjects not involved in the learning process are considered, the models discovered by our framework show higher generalization capability than those achieved without knowledge sharing: the improvement provided by knowledge sharing is equal to about 3.03% for precision, 1.56% for recall, 3.17% for F1, and 1.56% for accuracy. Moreover, statistical analysis reveals the statistical superiority of model exchange with respect to the case of no exchange taking place.

show abstract

Section: State Of the Artmentioning

confidence: 99%

A Federated Learning-Inspired Evolutionary Algorithm: Application to Glucose Prediction

Falco

Cioppa

Koutný

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Diabetes melitus (DM) merupakan salah satu penyakit mematikan di dunia [10]. Statistik meningkatkan bahwa, penyakit diabetes adalah salah satu penyakit paling berbahaya yang menyebabkan penyakit berbahaya lainnya dan dapat menyebabkan kematian dalam beberapa kasus [11]. Masalah diabetes melitus di Indonesia sudah terjadi sejak tahun 1980 sampai sekarang.…”

Section: Optimasi Algoritma K-nearest Neighbors Dengan Teknik Cross V...unclassified

Optimization of K-Nearest Neighbors Algorithm with Cross Validation Techniques for Diabetes Prediction with Streamlit

Prasetyo¹,

Laksana²

2022

JAIC

View full text Add to dashboard Cite

The problem that occurs in the application of K-Nearest Neighbors as a classification algorithm is the frequent occurrence of overfitting in data processing. This can be overcome by using cross-validation techniques in evaluating the algorithm model and minimizing overfitting. Then the performance of diabetes prediction accuracy is unknown using the K-Nearest Neighbors algorithm with cross-validation technique. The data used comes from the National Institute of Digestive and Kidney Diabetes in 2021. The case study in this study is to find out the initial screening for diabetes is supported by the results of algorithm accuracy and real time application of streamlit-based users. The purpose of this study was to optimize the accuracy results with a cross validation technique supported by the k-nearest neighbors algorithm in the study of diabetes data. The method used is the k-nearest neighbors algorithm which is supported by cross validation technique for optimal accuracy results. Then the application of a streamlit-based interactive web application for testing the accuracy results used by the user to see the probability that the user has diabetes. The results showed that the optimization of the Cross Validation technique supported by the KNearest Neighbors algorithm model worked well. The results of the confusion matrix using the cross validation technique are more accurate in terms of the advantages of using the cross-validation technique itself. So that the classification report which has a value of 95% is more accurate than the accuracy which is worth 92% because of the use of cross-validation techniques that can minimize overfitting in addition to considerations of the accuracy value and the implementation of streamlit-based interactive web applications for user testing is going well.

show abstract

“…Misalnya dalam penelitian dan pengolahan data penelitian, adanya data outlier akan menyebabkan bias pada hasil penelitian yang dilakukan [6]. Oleh karena itu, data outlier harus mendapatkan penanganan khusus, misalnya dilakukan transformasi terhadap data outlier tersebut dengan menurunkan value nya agar tidak terlalu jauh dengan kumpulan data yang ada, atau menghilangkan record tersebut apabila value nya terlalu jauh dari sebagian besar observasi yang ada [7]- [9].…”

Section: Pendahuluanunclassified

Analisis Metode DBSCAN (Density-Based Spatial Clustering of Application with Noise) dalam Mendeteksi Data Outlier

Armiady

2022

Jur. Ris. Kom.

View full text Add to dashboard Cite

Data outlier is data that is different from a group of data in a dataset. Data outlier will have an impact on the refraction of data analysis results, if not handled properly. Various approaches can be taken to detect data outlier, one of which is through the clustering method (grouping data). DBSCAN (Density-Based Spatial Clustering of Application with Noise) is a clustering method that is able to find data outlier in a data set. DBSCAN works by determining clusters based on data density, using the parameters epsilon (range) and MinPts (minimum points to form a cluster). This study aims to test several DBSCAN models that have different epsilon and MinPts parameters. The model used consists of 3 models, with details: Model 1 (eps=0,2, MinPts=5), Model 2 (eps=0,3, MinPts=5) and Model 3 (eps=0,4, MinPts =5). The dataset used is a dataset generated through the paint data feature on the Orange Data Mining tool, with 2 variables (x and y), with a total of 1051 data lines of records. The results obtained are that all the tested models found that there is 1 data point that is considered an outlier, namely the data is worth x = 0.370007 and y = 0.410475. In addition, from this research, it can also be concluded that the epsilon value affects the number of clusters formed. The higher the epsilo value, the smaller the number of clusters that may be formed

show abstract

Diabetes classification application with efficient missing and outliers data handling algorithms

Cited by 8 publications

References 28 publications

A Federated Learning-Inspired Evolutionary Algorithm: Application to Glucose Prediction

A Federated Learning-Inspired Evolutionary Algorithm: Application to Glucose Prediction

Optimization of K-Nearest Neighbors Algorithm with Cross Validation Techniques for Diabetes Prediction with Streamlit

Analisis Metode DBSCAN (Density-Based Spatial Clustering of Application with Noise) dalam Mendeteksi Data Outlier

Contact Info

Product

Resources

About