Dataset Information

Defining an Optimal Cut-Point Value in ROC Analysis: An Alternative Approach.

ABSTRACT: ROC curve analysis is often applied to measure the diagnostic accuracy of a biomarker. The analysis results in two gains: diagnostic accuracy of the biomarker and the optimal cut-point value. There are many methods proposed in the literature to obtain the optimal cut-point value. In this study, a new approach, alternative to these methods, is proposed. The proposed approach is based on the value of the area under the ROC curve. This method defines the optimal cut-point value as the value whose sensitivity and specificity are the closest to the value of the area under the ROC curve and the absolute value of the difference between the sensitivity and specificity values is minimum. This approach is very practical. In this study, the results of the proposed method are compared with those of the standard approaches, by using simulated data with different distribution and homogeneity conditions as well as a real data. According to the simulation results, the use of the proposed method is advised for finding the true cut-point.

SUBMITTER: Unal I

PROVIDER: S-EPMC5470053 | biostudies-literature | 2017

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Defining an Optimal Cut-Point Value in ROC Analysis: An Alternative Approach.

Unal Ilker I

Computational and mathematical methods in medicine 20170531

ROC curve analysis is often applied to measure the diagnostic accuracy of a biomarker. The analysis results in two gains: diagnostic accuracy of the biomarker and the optimal cut-point value. There are many methods proposed in the literature to obtain the optimal cut-point value. In this study, a new approach, alternative to these methods, is proposed. The proposed approach is based on the value of the area under the ROC curve. This method defines the optimal cut-point value as the value whose s ...[more]

PMID: 28642804

Similar Datasets

Project description:In this study we evaluated and redefined the optimum body mass index (BMI) cut-off point for the Iranian population based on metabolic syndrome (MeS) risk factors. We further evaluated BMI cut-off points with and without waist circumference (WC) as a cofactor of risk and compared the differences. This study is part of the largest surveillance programs conducted in Shiraz, Iran, termed the Shiraz Heart study. Our study sample included subjects between the ages of 20 to 65 years old. After excluding pregnant women, those with missing data and those with comorbid disease, a total of 12283 made up the study population. The participants underwent a series of tests and evaluations by trained professionals in accordance with WHO recommendations. Hypertension, abnormal fasting blood sugar (FBS), triglyceride (TG) and high density lipoprotein cholesterol (HDL) (in the context of the definition of metabolic syndrome) were prevalent among 32.4%, 27.6%, 42.1 and 44.2% of our participants, respectively. Women displayed higher rates of overall obesity compared to men (based on the definition by the WHO as higher than 30 kg/m2). Regarding MeS, 38.9% of our population had the all symptoms of MeS which was more prevalent among women (41.5% vs. 36%). When excluding WC in the definition of MeS, results showed that males tend to show a higher rate of metabolic risk factors (19.2% vs. 15.6%). Results of multivariate analysis showed that parallel to an increase in BMI, the odds ratio (OR) for acquiring each component of the metabolic syndrome increased (OR = 1.178; CI: 1.166-1.190). By excluding WC, the previous OR decreased (OR = 1.105; CI: 1.093-1.118). Receiver Operating Characteristic (ROC) curve analysis showed that the optimum BMI cut-off point for predicting metabolic syndrome was 26.1 kg/m2 and 26.2 kg/m2 [Accuracy (Acc) = 69% and 61%, respectively)] for males and females, respectively. The overall BMI cut-off for both sexes was 26.2 kg/m2 (Acc = 65%) with sensitivity and specificity of 69% and 62%, respectively. This cut-off had a positive predictive value of 54% and a negative predictive value of 76%. When we excluded waist circumference, the optimum BMI cut-off for acquiring metabolic risk factors in males decreased to 25.7 kg/m2 (Acc = 67%) and increased for women to 27.05 kg/m2 (Acc = 66%). Iranians are at higher risks of morbidity related to metabolic factors at a lower BMI cut-off and prompt action and preventive health policy are required to prevent and educate Iranians regarding diseases associated with obesity.

Project description:Equol is one of the most active soy isoflavones. When the association between soy food intake in daily life and health outcomes is examined in epidemiological studies, it is important to define the equol-producing status of each individual. However, few studies have assessed equol-producing status without a soy challenge test. To determine a robust cutoff criterion for equol producer classification in observational studies, we conducted a urinary isoflavone concentration survey in daily life among women. Furthermore, we examined the association between eating habits regarding soy foods and equol-producing status. A total of 4,412 participants were included in the analyses. Urinary isoflavones were analyzed using a high-performance liquid chromatography method. We examined the distribution of the log10 equol/daidzein ratios, finding a mixture of two normal distributions, corresponding to equol producer and non-producer subpopulations. Applying a finite mixture model, we estimated the means, standard deviations, and mixing proportions of these two distributions. The estimation was carried out using the SAS NLIN procedure. The optimal cutoff point for the log10 equol/daidzein ratio in the study population was determined to be -1.42, according to the estimated parameters of the mixture distribution. Based on this criterion, 1,830 (41.5%) of the participants were identified as equol producers. Compared with non-consumers of soy foods, consumers of soy foods had significantly higher odds of being equol producers. Using log10-transformed equol/daidzein ratios ? -1.42 to define equol producers among Japanese women is reasonable and suitable for determining equol-producing status in epidemiological studies. We found that soy food eating habits were associated with equol-producing status. Further investigation is required to evaluate associations between equol-producing status in daily life and health outcomes. The results of this study suggest the best cutoff point to use in the definition of equol-producing status in daily life.

Dataset Information

Defining an Optimal Cut-Point Value in ROC Analysis: An Alternative Approach.

Publications

Defining an Optimal Cut-Point Value in ROC Analysis: An Alternative Approach.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets