Article
Agriculture, Dairy & Animal Science
M. Martin, M. D. Kleinhenz, K. S. Schwartzkopf-Genswein, D. Melendez, S. Marti, E. A. Pajor, E. D. Janzen, J. Coetzee
Summary: This study used ROC analysis to evaluate the predictive value of pain biomarkers in livestock. It found that plasma cortisol, hair cortisol, and IRT had higher AUC values, while salivary cortisol, MNT, substance P, kinematic gait analysis, and a visual analog scale for pain had lower AUC values.
JOURNAL OF DAIRY SCIENCE
(2022)
Article
Computer Science, Hardware & Architecture
Ruhul Ali Khan
Summary: A new semiparametric model of the ROC curve is proposed, which is based on the resilience family or proportional reversed hazard family. The resulting ROC curve and its summary indices have simple analytic forms. The estimation methodologies of the resilience family and a simulation study to assess the performance of the estimators are discussed.
IEEE TRANSACTIONS ON RELIABILITY
(2023)
Article
Health Care Sciences & Services
Ruitao Lin, K. C. Gary Chan, Haolun Shi
Summary: The study proposes a method for exact inference using empirical likelihood within a Bayesian framework to estimate the area under the receiver operating characteristic curve, drawing inference from posterior samples obtained via a Gibbs sampler. The method simplifies computation and can be applied to various scenarios in different fields.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2021)
Article
Mathematics, Applied
Alexej Gossmann, Aria Pezeshk, Yu-Ping Wang, Berkman Sahiner
Summary: The performance evaluation of constantly evolving machine learning algorithms, especially in high-risk fields such as medicine, faces new challenges. Reusing the same test dataset can lead to overfitting and overly optimistic conclusions about algorithm performance. A modified holdout mechanism shows potential in reducing overfitting issues.
SIAM JOURNAL ON MATHEMATICS OF DATA SCIENCE
(2021)
Article
Biology
M. Amico, I Van Keilegom, B. Han
Summary: This article introduces a cure model for handling cure fractions in survival data, and proposes a ROC curve estimator to evaluate the classification performance of cured/noncured status. Through simulations and comparisons, the good performance of the proposed method is demonstrated, applied to a breast cancer dataset.
Article
Health Care Sciences & Services
Pablo Martinez-Camblor
Summary: A good diagnostic test should exhibit distinct behavior for positive and negative populations, but that is not sufficient to create a good classification system. Defining decision criteria is crucial for solving the complex binary classification problem. The receiver-operating characteristic curve length serves as an index for the optimal discriminatory capacity of a biomarker, but its estimation requires parametric or smoothed models. This study examines a kernel density estimator-based approximation for estimating the length of the receiver-operating characteristic curve and explores its behavior through statistical analysis and Monte Carlo simulations with real-world examples.
STATISTICAL METHODS IN MEDICAL RESEARCH
(2023)
Article
Ecology
Wenkai Li, Qinghua Guo
Summary: The performance evaluation of species distribution models is typically done through receiver operating characteristic (ROC) and precision-recall (PR) plots. This study introduces a new approach (PB approach) to calibrate ROC/PR curves from presence and background data, which provides highly accurate estimations in experiments.
ECOLOGY AND EVOLUTION
(2021)
Article
Environmental Sciences
Dal Rae Jin, Mikyung Lee, Hae Jong Yang, Shin Kim, Jung-Suk Lee, Seong-Dae Moon
Summary: This study evaluates metal contamination in brackish areas of Korea using both freshwater-sediment quality guidelines and marine-sediment quality guidelines. The results indicate that freshwater-sediment quality guidelines are more suitable for evaluating brackish sediments in South Korea.
MARINE POLLUTION BULLETIN
(2022)
Article
Oncology
Takeshi Hashimoto, Osamu Komori, Jun Nakashima, Takeshi Kashima, Yuri Yamaguchi, Naoya Satake, Yoshihiro Nakagami, Toshihide Shishido, Kazunori Namiki, Yoshio Ohno
Summary: This study aimed to establish a novel method of using a PSA threshold nomogram to predict pathologically advanced prostate cancer, and developing an individualized PSA threshold nomogram using AUCBoost statistical method, which may be useful in determining treatment strategies before surgery. The study analyzed the medical records of patients with clinically localized prostate cancer and identified clinical covariates significantly associated with pathological tumor stage. The results showed that a combination model comprising PSA, prostate volume, clinical tumor stage, percent positive core, and Gleason Grade Group produced the highest AUC for predicting pathological tumor stage.
UROLOGIC ONCOLOGY-SEMINARS AND ORIGINAL INVESTIGATIONS
(2022)
Article
Operations Research & Management Science
I. Edhem Sakarya, O. Erhun Kundakcioglu
Summary: The purpose of this study is to solve the multi-instance classification problem by maximizing the area under the Receiver Operating Characteristic (ROC) curve obtained for witness instances. A mixed integer linear programming model is derived to select witnesses and produce the best possible ROC curve using a linear ranking function. The study also explores generating new features for tackling non-linear separable data and conducts a comprehensive computational study to compare the proposed methods with state-of-the-art approaches.
JOURNAL OF GLOBAL OPTIMIZATION
(2023)
Article
Mathematical & Computational Biology
Vanda Inacio, Vanda M. Lourenco, Miguel de Carvalho, Richard A. Parker, Vincent Gnanapragasam
Summary: This study develops a robust and flexible model for inference on covariate-specific ROC curves, aiming to safeguard against the impact of outlying test results and accommodate for nonlinear effects of covariates. Simulation results demonstrate the method successfully recovers the true covariate-specific area under the ROC curve in various contamination scenarios.
STATISTICS IN MEDICINE
(2021)
Article
Mathematical & Computational Biology
Pablo Martinez-Camblor, Sonia Perez-Fernandez, Susana Diaz-Coto
Summary: ROC curve and AUC are commonly used for evaluating the discriminatory ability of markers, while gROC curve and gAUC are extensions to handle more complex situations and provide a more comprehensive assessment of classification capacity. This paper delves into the properties of gAUC, providing expressions for asymptotic variance and covariance, as well as proposing a non-parametric procedure for comparison. Monte Carlo simulations and a real-world example are used to illustrate the practical application of these methods.
INTERNATIONAL JOURNAL OF BIOSTATISTICS
(2022)
Article
Statistics & Probability
Alicja Jokiel-Rokita, Rafal Topolnicki
Summary: This paper introduces the use of the minimum distance method to estimate a one-dimensional parameter of an ROC curve based on the Lehmann family of distributions, proving the consistency and asymptotic normality of the obtained estimators. Through a simulation study, the accuracy of the proposed estimators is compared with that of known estimators based on different methods, with an illustrative example of an application to real data provided.
Article
Psychology, Multidisciplinary
Yueran Yang
Summary: This article discusses how to improve classification performance using ROC analysis and introduces a tool to visualize a classifier's expected utility. The analysis reveals that expected utility depends not only on the accuracy of a classifier but also on its operating point. Therefore, choosing the optimal operating point can maximize expected utility. The article also explores other methods beyond ROC analysis to increase expected utility.
PSYCHOLOGICAL METHODS
(2022)
Article
Geochemistry & Geophysics
Chein-I Chang
Summary: This article explores fundamental and conceptual issues in the application of the 2D ROC curve for hyperspectral anomaly detection, providing solutions and insights. By deriving a mathematical theory and conducting comprehensive analysis, it reveals the principles of plotting the 2D ROC curve and evaluating background suppression. The article also highlights that many detectors claiming good performance in AD actually perform poorly in BS.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)