• International Journal of Internet, Broadcasting and Communication
• /
• v.16 no.4
• /
• pp.290-300
• /
• 2024

This study presents a comprehensive evaluation of various machine learning models for predicting heart failure outcomes. Leveraging a data set of clinical records, the performance of Logistic Regression, Support Vector Machine (SVM), Random Forest, Soft Voting ensemble, and XGBoost models are rigorously assessed using multiple evaluation metrics, including accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC). The analysis reveals that the XGBoost model outperforms the other techniques across all metrics, exhibiting the highest AUC score, indicating superior discriminative ability in distinguishing between patients with and without heart failure. Furthermore, the study highlights the importance of feature importance analysis provided by XGBoost, offering valuable insights into the most influential predictors of heart failure, which can inform clinical decision-making and patient management strategies. The research also underscores the significance of balancing precision and recall, as reflected by the F1-score, in medical applications to minimize the consequences of false negatives.

• The Korean Journal of Health Service Management
• /
• v.11 no.2
• /
• pp.105-115
• /
• 2017

Objectives : The aim of this study was to suggest the optimal cut off for best, very good, good, slightly bad, and bad grades. Methods : The subjects were recruited from 4 areas of South Korea and 487 questionnaires were analyzed. The nominal and continuous self-rated health questions were used to reveal the optimal cut off and the Short Form-12 Health Survey questionnaire (SF-12) was additionally used. Frequency, Pearson's correlation coefficient, and ROC-curve analysis were used; the significance level was <.05. Results : Subjects assigned 15(3.1%), 90(18.5%), 237(48.7%), 130(26.7%), and 15(3.1%) to best, very good, good, slightly bad and bad groups respectively. The self-rated health score was associated with total Component (r=.563, p<.001), Physical Component (r=.520, p<.001) and Mental Component of SF-12 (r=.303, p<.001). The optimal cut off was 80.5, 70.5, 53.5, and 40.5 for best, very good or more, good or more, and under slightly bad respectively and area under curve was 0.898, 0.908, 0.945, and 0.908 accordingly. Conclusions : This study suggests that the self-rated health score and grade could be integrated with the optimal cut off.

• Journal of Physiology & Pathology in Korean Medicine
• /
• v.32 no.5
• /
• pp.341-346
• /
• 2018

The aims of this study were to evaluate the reliability and validity of the cold and heat pattern identification questionnaire (CHPIQ). From July 2015 to December 2015, 120 participants, university faculties, filled out CHPIQ by the way of self-reporting. Then two Korean medical doctors independently diagnosed them whether they belonged to cold pattern (CP) or not, and heat pattern (HP) or not. We evaluated the internal consistency using Cronbach's alpha coefficient, and the validity using the sensitivity and specificity through receiver operating characteristic-curve. The internal consistency (Cronbach's alpha coefficient) showed 0.754 (CP) and 0.753 (HP). The area under the curve was recorded with 0.884 (CP) and 0.786 (HP). The agreements between CHPIQ and experts were 82.8% (CP) and 72.9% (HP). The sensitivities showed 0.707 (CP) and 0.719 (HP), and the specificities were 0.935 (CP) and 0.736 (HP). This study suggests that CHPIQ is a reliable and valid instrument for estimating cold-heat pattern identification.

• The Korean Journal of Applied Statistics
• /
• v.31 no.3
• /
• pp.329-342
• /
• 2018

Nomogram is a statistical tool that visualizes the risk factors of the disease and then helps to understand the untrained people. This study used risk factors of chronic obstructive pulmonary disease (COPD) and compared with logistic regression model and naïve Bayesian classifier model. Data were analyzed using the Korean National Health and Nutrition Examination Survey 6th (2013-2015). First, we used 6 risk factors about COPD. We constructed nomogram using logistic regression model and naïve Bayesian classifier model. We also compared the nomograms constructed using the two methods to find out which method is more appropriate. The receiver operating characteristic curve and the calibration plot were used to verify each nomograms.

• Communications for Statistical Applications and Methods
• /
• v.17 no.2
• /
• pp.275-292
• /
• 2010

Receiver operating characteristic(ROC) curves can be used to assess the accuracy of tests measured on ordinal or continuous scales. The most commonly used measure for the overall diagnostic accuracy of diagnostic tests is the area under the ROC curve(AUC). When two ROC curves are constructed based on two tests performed on the same individuals, statistical analysis on differences between AUCs must take into account the correlated nature of the data. This article focuses on confidence interval estimation of the difference between paired AUCs. We compare nonparametric, maximum likelihood, bootstrap and generalized pivotal quantity methods, and conduct a monte carlo simulation to investigate the probability coverage and expected length of the four methods.

• International Journal of Computer Science & Network Security
• /
• v.21 no.9
• /
• pp.179-184
• /
• 2021

In our study we performed a sentiments analysis from the images. For this purpose, we used 153 images that contain: people, animals, buildings, landscapes, cakes and objects that we divided into two categories: images that suggesting a positive or a negative emotion. In order to classify the images using the two categories, we created two models. The SAI-G model was created with Google's AutoML Vision service. The SAI-C model was created on the Clarifai platform. The data were labeled in a preprocessing stage, and for the SAI-C model we created the concepts POSITIVE (POZITIV) AND NEGATIVE (NEGATIV). In order to evaluate the performances of the two models, we used a series of evaluation metrics such as: Precision, Recall, ROC (Receiver Operating Characteristic) curve, Precision-Recall curve, Confusion Matrix, Accuracy Score and Average precision. Precision and Recall for the SAI-G model is 0.875, at a confidence threshold of 0.5, while for the SAI-C model we obtained much lower scores, respectively Precision = 0.727 and Recall = 0.571 for the same confidence threshold. The results indicate a lower classification performance of the SAI-C model compared to the SAI-G model. The exception is the value of Precision for the POSITIVE concept, which is 1,000.

• Journal of Korean Academy of Nursing
• /
• v.50 no.6
• /
• pp.787-798
• /
• 2020

Purpose: This study aimed to assign weights for subscales and items of the Post-Intensive Care Syndrome questionnaire and suggest optimal cut-off values for screening unplanned hospital readmissions of critical care survivors. Methods: Seventeen experts participated in an analytic hierarchy process for weight assignment. Participants for cut-off analysis were 240 survivors who had been admitted to intensive care units for more than 48 hours in three cities in Korea. We assessed participants using the 18-item Post-Intensive Care Syndrome questionnaire, generated receiver operating characteristic curves, and analysed cut-off values for unplanned readmission based on sensitivity, specificity, and positive likelihood ratios. Results: Cognitive, physical, and mental subscale weights were 1.13, 0.95, and 0.92, respectively. Incidence of unplanned readmission was 25.4%. Optimal cut-off values were 23.00 for raw scores and 23.73 for weighted scores (total score 54.00), with an area of under the curve (AUC) of .933 and .929, respectively. There was no significant difference in accuracy for original and weighted scores. Conclusion: The optimal cut-off value accuracy is excellent for screening of unplanned readmissions. We recommend that nurses use the Post-Intensive Care Syndrome Questionnaire to screen for readmission risk or evaluating relevant interventions for critical care survivors.

• Clinical and Experimental Reproductive Medicine
• /
• v.48 no.2
• /
• pp.156-162
• /
• 2021

Objective: The aim of the present study was to evaluate the predictive capability of fasting-state measurements of glucose and insulin levels alone for abnormal glucose tolerance in women with polycystic ovary syndrome (PCOS). Methods: In total, 153 Korean women with PCOS were included in this study. The correlations between the 2-hour postload glucose (2-hr PG) level during the 75-g oral glucose tolerance test (OGTT) and other parameters were evaluated using Pearson correlation coefficients and linear regression analysis. The predictive accuracy of fasting glucose and insulin levels and other fasting-state indices for assessing insulin sensitivity derived from glucose and insulin levels for abnormal glucose tolerance was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Significant correlations were observed between the 2-hr PG level and most fasting-state parameters in women with PCOS. However, the area under the ROC curve values for each fasting-state parameter for predicting abnormal glucose tolerance were all between 0.5 and 0.7 in the study participants, which falls into the "less accurate" category for prediction. Conclusion: Fasting-state measurements of glucose and insulin alone are not enough to predict abnormal glucose tolerance in women with PCOS. A standard OGTT is needed to screen for impaired glucose tolerance and type 2 diabetes mellitus in women with PCOS.

• Imaging Science in Dentistry
• /
• v.52 no.2
• /
• pp.187-195
• /
• 2022

Purpose: This study developed a convolutional neural network (CNN) model to diagnose maxillary sinusitis on panoramic radiographs(PRs) and cone-beam computed tomographic (CBCT) images and evaluated its performance. Materials and Methods: A CNN model, which is an artificial intelligence method, was utilized. The model was trained and tested by applying 5-fold cross-validation to a dataset of 148 healthy and 148 inflamed sinus images. The CNN model was implemented using the PyTorch library of the Python programming language. A receiver operating characteristic curve was plotted, and the area under the curve, accuracy, sensitivity, specificity, positive predictive value, and negative predictive values for both imaging techniques were calculated to evaluate the model. Results: The average accuracy, sensitivity, and specificity of the model in diagnosing sinusitis from PRs were 75.7%, 75.7%, and 75.7%, respectively. The accuracy, sensitivity, and specificity of the deep-learning system in diagnosing sinusitis from CBCT images were 99.7%, 100%, and 99.3%, respectively. Conclusion: The diagnostic performance of the CNN for maxillary sinusitis from PRs was moderately high, whereas it was clearly higher with CBCT images. Three-dimensional images are accepted as the "gold standard" for diagnosis; therefore, this was not an unexpected result. Based on these results, deep-learning systems could be used as an effective guide in assisting with diagnoses, especially for less experienced practitioners.

• Annual Conference of KIPS
• /
• 2021.11a
• /
• pp.805-807
• /
• 2021

In-hospital cardiac arrest is a significant problem for medical systems. Although the traditional early warning systems have been widely applied, they still contain many drawbacks, such as the high false warning rate and low sensitivity. This paper proposed a strategy that involves a deep learning approach based on a novel interpretable deep tabular data learning architecture, named TabNet, for the Rapid Response System. This study has been processed and validated on a dataset collected from two hospitals of Chonnam National University, Korea, in over 10 years. The learning metrics used for the experiment are the area under the receiver operating characteristic curve score (AUROC) and the area under the precision-recall curve score (AUPRC). The experiment on a large real-time dataset shows that our method improves compared to other machine learning-based approaches.