Unveiling diabetes onset: Optimized XGBoost with Bayesian optimization for enhanced prediction

PLoS ONE, Год журнала: 2025, Номер 20(1), С. e0310218 - e0310218

Опубликована: Янв. 24, 2025

Diabetes, a chronic condition affecting millions worldwide, necessitates early intervention to prevent severe complications. While accurately predicting diabetes onset or progression remains challenging due complex and imbalanced datasets, recent advancements in machine learning offer potential solutions. Traditional prediction models, often limited by default parameters, have been superseded more sophisticated approaches. Leveraging Bayesian optimization fine-tune XGBoost, researchers can harness the power of data analysis improve predictive accuracy. By identifying key factors influencing risk, personalized prevention strategies be developed, ultimately enhancing patient outcomes. Successful implementation requires meticulous management, stringent ethical considerations, seamless integration into healthcare systems. This study focused on optimizing hyperparameters an XGBoost ensemble model using optimization. Compared grid search (accuracy: 97.24%, F1-score: 95.72%, MCC: 81.02%), with achieved slightly improved performance 97.26%, MCC:81.18%). Although improvements observed this are modest, optimized represents promising step towards revolutionizing treatment. approach holds significant outcomes for individuals at risk developing diabetes.

Язык: Английский

---

Boosting the Accuracy of Cardiovascular Disease Prediction Through SMOTE

Опубликована: Июнь 23, 2023

Cardiovascular Disease (CVD) affects deaths and hospitalisations. Clinical data analytics struggles to predict heart disease survival. This report compares machine learning-based cardiovascular prediction studies. The authors use a Kaggle dataset of 70,000 records 16 features show SMOTE model with hyperparameter-optimized classifiers. Random Forest outperforms KNN 13 elements in prediction. Naive Bayes SVM on complete feature sets. proposed achieves 86% accuracy, the optimised technique traditional all metrics. study analyses strengths weaknesses existing models for making predictions learning suggests promising new method.

Язык: Английский

---

Interpretable Machine Learning Framework to Predict the Glass Transition Temperature of Polymers

Polymers, Год журнала: 2024, Номер 16(8), С. 1049 - 1049

Опубликована: Апрель 10, 2024

The glass transition temperature of polymers is a key parameter in meeting the application requirements for energy absorption. Previous studies have provided some data from slow, expensive trial-and-error procedures. By recognizing these data, machine learning algorithms are able to extract valuable knowledge and disclose essential insights. In this study, dataset 7174 samples was utilized. were numerically represented using two methods: Morgan fingerprint molecular descriptor. During preprocessing, scaled standard scaler technique. We removed features with small variance used Pearson correlation technique exclude that highly connected. Then, most significant selected recursive feature elimination method. Nine techniques employed predict tune their hyperparameters. models compared performance metrics mean absolute error (MAE), root square (RMSE), coefficient determination (R2). observed extra tree regressor best results. Significant also identified statistical methods. SHAP method demonstrate influence each on model's output. This framework can be adaptable other properties at low computational expense.

Язык: Английский

---

Antidiabetic Potential of Senna siamea: α‐Glucosidase Inhibition, Postprandial Blood Glucose Reduction, Toxicity Evaluation, and Molecular Docking

Scientifica, Год журнала: 2025, Номер 2025(1)

Опубликована: Янв. 1, 2025

Senna siamea (Lam.) H.S. Irwin & Barneby is used in Thai cuisine. This plant also traditional treatments, including diabetes. Therefore, this study aims to examine the antihyperglycemic effects of S. heartwood extract. The ethanolic extract exhibited activity against α ‐glucosidase enzyme with IC 50 values 54.4 μg/mL. Moreover, (250–1000 mg/kg BW) was tested using normal rats and without sucrose 3 g/kg BW administration. results showed that all concentrations significantly reduced fasting blood glucose compared control. In addition, agreed amount small intestine rats. acute toxicity study, a single dose at 2000 caused no mortality, hematological biochemical parameters revealed toxic on subchronic administration for 90 days, 250 BW, significant changes treated groups control group. However, histopathology liver kidney indicated an inflammatory response 500 1000 extract, correlating findings. Finally, molecular docking conducted evaluate theoretical interactions between three main stilbenes previously found mammalian ‐glucosidases (Wistar rat human). simulation supported vivo suggested potential human glucosidase inhibition. could be promising candidate ‐glucosidase. offers encouraging information natural compounds from act as inhibitors diabetes treatment through drug development or dietary supplement hyperglycemia individuals.

Язык: Английский

---

An Innovative Ensemble Deep Learning Clinical Decision Support System for Diabetes Prediction

IEEE Access, Год журнала: 2024, Номер 12, С. 106193 - 106210

Опубликована: Янв. 1, 2024

Diabetes is a significant global health concern, with an increasing number of diabetic people at risk. It considered chronic disease and leads to fatalities annually. Early prediction diabetes essential for preventing its progression reducing the risk severe complications such as kidney heart diseases. This study proposes innovative Ensemble Deep Learning (EDL) clinical decision support system high accuracy. The proposed EDL model uses (DL) architectures Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM), Convolutional (CNN), integrated ensemble learning-based stacking model. implemented based on stack that applies meta-level models, including stack-ANN, stack-CNN, stack-LSTM, improve diabetes. Three datasets, I. Pima Indian Dataset (PIMA-IDD-I), II. Frankfurt Hospital Germany (DDFH-G), III. Iraqi Patient (IDPD-I) are used train novel models. Extra Tree Classifier (ETC) approach extract relevant features from data. performance models evaluated major evaluation metrics accuracy, precision, sensitivity, specificity, F-score, Matthews Correlation Coefficient (MCC), ROC/AUC. Among stack-ANN achieved robust using DDFH-G, PIMA-IDD-I, IDPD-I datasets accuracy scores 99.51%, 98.81%, 98.45%, respectively. overall results demonstrate outperform previous studies in predicting

Язык: Английский

---

Recent trends and perspectives of artificial intelligence-based machine learning from discovery to manufacturing in biopharmaceutical industry

Journal of Pharmaceutical Investigation, Год журнала: 2023, Номер 53(6), С. 803 - 826

Опубликована: Ноя. 1, 2023

Язык: Английский

---

Adapted Deep Ensemble Learning-Based Voting Classifier for Osteosarcoma Cancer Classification

Diagnostics, Год журнала: 2023, Номер 13(19), С. 3155 - 3155

Опубликована: Окт. 9, 2023

The study utilizes osteosarcoma hematoxylin and the Eosin-stained image dataset, which is unevenly dispersed, it raises concerns about potential impact on overall performance reliability of any analyses or models derived from dataset. In this study, a deep-learning-based convolution neural network (CNN) adapted heterogeneous ensemble-learning-based voting classifier have been proposed to classify osteosarcoma. methods can also resolve issue develop unbiased learning by introducing an evenly distributed training Data augmentation employed boost generalization abilities. Six different pre-trained CNN models, namely MobileNetV1, Mo-bileNetV2, ResNetV250, InceptionV2, EfficientNetV2B0, NasNetMobile, are applied evaluated in frozen fine-tuned-based phases. addition, novel model developed model, fine-tuned NasNetMobile Efficient-NetV2B0 introduced outperforms other models. Kappa score obtained 93.09%. Notably, attains highest 96.50% all findings practical implications telemedicine, mobile healthcare systems, as supportive tool for medical professionals.

Язык: Английский

---

Enhancing Diabetes Prediction and Prevention through Mahalanobis Distance and Machine Learning Integration

Applied Sciences, Год журнала: 2024, Номер 14(17), С. 7480 - 7480

Опубликована: Авг. 23, 2024

Diabetes mellitus (DM) is a global health challenge that requires advanced strategies for its early detection and prevention. This study evaluates the South Korean population using Korea National Health Nutrition Examination Survey (KNHANES) dataset from 2015 to 2021, provided by Disease Control Prevention Agency (KDCA), focusing on improving diabetes prediction models. Outlier removal was implemented Mahalanobis distance (MAH), feature selection based multicollinearity (MC) reliability analysis (RA). The proposed Extreme Gradient Boosting (XGBoost) model demonstrated exceptional performance, achieving an accuracy of 98.04% (95% CI: 97.89~98.59), F1-score 98.24%, Area Under Curve (AUC) 98.71%, outperforming other state-of-the-art highlights significance rigorous outlier in enhancing predictive power risk Notably, significant increase cases observed during COVID-19 pandemic, particularly linked male sex, older age, rural location, hypertension, obesity, underscoring need enhanced public intervention targeted

Язык: Английский

---

Corporate genome screening India (CoGsI) identified genetic variants association with T2D in young Indian professionals

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 2, 2025

Abstract Rising cases of type 2 diabetes (T2D) in India, especially metropolitan cities is an increasing concern. The individuals that were most affected are young professionals working the corporate sector. However, sector has remained least explored for T2D risk predisposition. Considering employees’ lifestyles and role gene-environment interaction susceptibility, study aims to find genetic variants associated with In this first kind study, 680 (284 cases, 396 controls) diagnosed screened 2658 on array designed explicitly CoGsI study. variant filtering was done at Bonferroni p-value 0.000028. data analysed using PLINK v1.09, SPSS, R programming, VEP tool, FUMA GWAS tool. Interestingly, 42 risk. Out 42, three missense (rs1402467, rs6050, rs713598) Sulfotransferase family 1 C member 4 ( SULT1C4 ), Fibrinogen Alpha Chain FGA Taste Receptor Member 38 TAS2R38 ) two untranslated region (UTR) (rs1063320 rs6296) Major Histocompatibility Complex, Class I, G HLA-G 5-Hydroxytryptamine 1B HTR1B identified potential genomic markers susceptibility early onset T2D. Present findings provide insights into mechanisms underlying manifestation due genetics interacting occupational stress urban lifestyles.

Язык: Английский

---

Hybrid Machine Learning Models for Accurate Type 2 Diabetes Mellitus Prediction Using a Stacking Classifier and a Meta-Model Approach

Cureus Journal of Computer Science., Год журнала: 2025, Номер unknown

Опубликована: Март 3, 2025

Язык: Английский

---