Receiver Operating Characteristic Curve Method Research Articles

Flood susceptibility modeling by integrating tree‐based regression with metaheuristic algorithm, BWO

AbstractFloods are becoming more widely acknowledged as a common occurrence of nature's dangers on a global scale. Although forecasting models primarily focus on timely warnings, models aimed at evaluating dangerous zones can play a vital role in shaping policies for adaptation, mitigation, and reducing the risk of disasters. Using machine learning techniques including hybrid black widow optimization (BWO) with XGBoost, LGBoost, and AdaBoost. We generate a flood susceptibility map for considered region of lower mahanadi basin (LMB). This study examines the effectiveness of these machine learning models in assessing and mapping flood susceptibility, while also providing suggestions for future research in this area. Flood susceptibility model was developed using 13 variables: Altitude, Aspect, Curvature, Distance from river, Drainage Density, Stream Power Index (SPI), Sediment Transport Index (STI), Rainfall intensity, Land Use Land Cover (LULC), Topographic Wetness Index (TWI), Terrain Roughness Index (TRI), Normalized Difference Vegetation Index (NDVI), and slope. Additionally, flood inventory data were incorporated into the model. Dataset was divided into a 70% portion for training model and a 30% portion for validating model. To assess the performance of the model, several evaluation metrics were employed, including receiver operating characteristic (ROC) curve and other performance indices. Evaluation of flood susceptibility mapping, using ROC curve method in combination with flood density yielded strong and reliable results for various models. BWO‐XGBoost achieved a score of 0.889, BWO‐LGBoost achieved a score of 0.937, and BWO‐ADABoost achieved a score of 0.904. These scores indicate effectiveness of these models in accurately predicting flood susceptibility in the study area. A comparison was made with commonly used methods in flood susceptibility assessment to evaluate the efficiency of proposed models. It was found that having a first‐class and enlightening database is crucial for accurately classifying flood types in flood susceptibility mapping. This aspect greatly contributes to improving the overall performance of the model. Among the evaluated methods, the hybrid model BWO‐LGBoost demonstrated better performance compared with others, indicating its effectiveness in accurately predicting flood susceptibility.

Aquifers Recharge Potentiality Evaluation using Bivariate and Multivariate Statistical Analysis: A Case Study in Malwathu Oya River Basin

An aquifer is a body of porous rock or sediment saturated with groundwater that plays a vital role in the hydrological cycle. Due to rapid population growth and urbanization, these valuable resources face threats like climate changes, pollution, and over-exploitation. Overdraft, where extraction exceeds recharge, can lead to groundwater depletion. Although Sri Lanka is not considered a water- scarce country, it faces rising water demands due to rapid population growth. Especially in dry zones, groundwater is crucial for drinking and agriculture; unauthorized agro-wells exacerbate groundwater issues. Therefore, this research aims to assess groundwater recharge potential zones in Malwathu Oya river basin using Bivariate and Multivariate statistical analysis. Comparing these techniques will lead to the precise identification of potential recharge areas. And this study primarily focuses on the implementation of GIS to find solutions for this kind of challenge without performing field investigations. Therefore, eight conditioning factors; water level, electrical conductivity, lithology, soil, slope, lineament density, drainage density and land use were selected as groundwater recharge influencing factors. The well inventory database consisted of 380 well locations and, out of these, 70% was considered as the training dataset, and the remaining 30% was taken as the testing data set. Groundwater recharge potential maps (GRP) were developed in ArcGIS 10.8, based on the results obtained from Frequency Ratio (FR), Logistic Regression (LR) and their ensemble method. For testing the accuracy of the FR model, test well locations were cross tabulated with the GRP map, and the highest well percentage was achieved in high groundwater recharge potential zone (37.39%). In the LR model, cross-tabulation results showed that 62.61% of testing wells were in the very high potential zones. The ensemble method of FR and LR revealed that 53.04% of test wells fall into the very high potential class. Hence, it can be assumed that LR predictions are more reliable than the FR model. To further validate the LR model and ensemble model, Receiver Operating Characteristics curve method was used. The Area Under the Curve (AUC) values for Success rate curve (SRC) and prediction rate curve (PRC) in LR method were 63% and 57% respectively. In ensemble method, the AUC values for SRC and PRC were 89% and 75%. These results indicate that the LR model is relatively poor estimator, whereas FR and LR ensemble method is relatively good estimator to evaluate groundwater recharge potential in Mulwathu Oya river basin.
 
 Keywords: Area under curve, Frequency ratio, GIS, Groundwater recharge, Logistic regression

Forest Fire Risk Mapping Using GIS Based Analytical Hierarchy Process Approach

Turkiye is located in a region sensitive to forest fires due to the Mediterranean climate, vegetation and topography of the south and west coasts. Every year, forest fires for various reasons cause thousands of hectares of forest area to be burned. Fires damage the ecosystem as well as have economic consequences. While the Mediterranean, Aegean and Marmara coasts, where the Mediterranean climate and fire-sensitive tree species are dominant, are at primary risk against forest fires, the forests located on the Black Sea coasts where the Black Sea climate is dominant are less risky in terms of fire. For an effective fight against forest fires, it is crucial to identify zones with fire risk based on various parameters such as forest structures (tree species, crown closure, stand age), topographic features (slope, aspect), climate, and proximity to certain points (such as roads, settlements, agricultural areas). Fire risk data will shed light on the measures that can be taken against fire. In this study, GIS (Geographic Information Systems) based Analytical Hierarchy Process (AHP) method, one of the well know MCDA (Multi-criteria Decision Making Analysis) methods, was used to develop the fire risk map of Mersin Forestry Regional Directorate (FRD) within the Mediterranean region of Turkiye. Then, the accuracy of the fire risk map was evaluated by taking into account the previous fires in the regional directorate. As a result, it was found that 13.87% of the study area was classified as very high, 25.87% high, 24.68% medium, 22.44% low and 13.14% very low risk areas. The results showed that tree species are the most effective risk factor in forest fires, and followed by stand age factor. The accuracy of the fire risk map was evaluated by using the location information of a total of 562 fires in Mersin FRD between 2003-2022 that damaged forest areas. In order to determine the accuracy of the fire risk map, ROC (Receiver Operating Characteristic) curve method was used in ArcGIS environment. As a result, the AUC (Area Under Curve) value was found to be approximately 74%, and these values showed that the fire risk map developed for Mersin FRD was moderately reliable. With this study, it has been demonstrated that it is possible to produce reliable fire risk maps in a short time using the GIS-based AHP method.

Basic assessment on adding platelet measurement to legal health checkup in Japan: A cross-sectional and 20-year longitudinal study.

In Japan, health checkups for workers are legally compulsory. Considering legal health checkup items are important for Japanese workers' health problems. To date, the legal health checkup items for blood cell counts include only red blood cell counts and hemoglobin but not platelet counts. This study aimed to investigate the significance of measuring platelets in workers by showing the association between the FIB-4 index (FIB-4), which can be easily calculated from factors including platelet counts and viral hepatitis infection. Both cross-sectional and longitudinal analyses were conducted on the comprehensive medical examinations of male workers. In fiscal year (FY) 2019, a logistic regression model was applied to 12,918 examinees. For 13,459 examinees (mean age = 47.5 ± 9.3 SD), FY2000 was set to be followed until FY2019. A total of 149,956 records between FY2000 and FY2019 were analyzed cross-sectionally, and 8,038 men who were consecutively examined to FY2019 at the longest were analyzed longitudinally. Receiver operating characteristic (ROC) curve-area under the ROC curve (ROC-AUC) and Cox proportional methods were used to examine the association between platelet-related indices and viral hepatitis infection. Logistic regression showed that the risk of FIB-4 ≥ 2.67 was mostly associated with hepatitis C virus antibody (HCVAb) positivity [odds ratio (OR) = 2.51, 95% confidence interval (CI) = 1.08-5.86], while negatively associated with body mass index (BMI) (OR = 0.54, 95% CI = 0.30-0.97), and not associated with the presence of fatty liver. To detect HVC Ab positivity, ROC-AUC showed more effectiveness in FIB-4 than in the AST/ALT ratio (0.776, 95% CI = 0.747-0.773 vs. 0.552; 95% CI = 0.543-0.561). The Cox analysis showed that the risk of FIB-4 ≥ 2.67 was closely associated with hepatitis B virus surface antigen (HBsAg) [hazard ratio (HR) = 3.1, 95% CI = 2.0-4.6] and HCV Ab positivity (HR = 3.2, 95% CI = 2.0-5.0). Our results suggest that it might be worth considering that usage of information on platelets in legal health checkups could be some help not to overlook workers with hepatitis virus carriers as a complementary countermeasure, although further investigations are needed into its practical application.

Study on the sensitivity of a volumetric modulated arc therapy plan verification equipment on multi-leaf collimator opening and closing errors and its gamma pass rate limit

To investigate the γ pass rate limit of plan verification equipment for volumetric modulated arc therapy (VMAT) plan verification and its sensitivity on the opening and closing errors of multi-leaf collimator (MLC), 50 cases of nasopharyngeal carcinoma VMAT plan with clockwise and counterclockwise full arcs were randomly selected. Eight kinds of MLC opening and closing errors were introduced in 10 cases of them, and 80 plans with errors were generated. Firstly, the plan verification was conducted in the form of field-by-field measurement and true composite measurement. The γ analysis with the criteria of 3% dose difference, distance to agreement of 2 mm, 10% dose threshold, and absolute dose global normalized conditions were performed for these fields. Then gradient analysis was used to investigate the sensitivity of field-by-field measurement and true composite measurement on MLC opening and closing errors, and the receiver operating characteristic curve (ROC) was used to investigate the optimal threshold of γ pass rate for identifying errors. Tolerance limits and action limits for γ pass rates were calculated using statistical process control (SPC) method for another 40 cases. The error identification ability using the tolerance limit calculated by SPC method and the universal tolerance limit (95%) were compared with using the optimal threshold of ROC. The results show that for the true composite measurement, the clockwise arc and the counterclockwise arc, the descent gradients of the γ passing rate with per millimeter MLC opening error are 10.61%, 7.62% and 6.66%, respectively, and the descent gradients with per millimeter MLC closing error are 9.75%, 7.36% and 6.37%, respectively. The optimal thresholds obtained by the ROC method are 99.35%, 97.95% and 98.25%, respectively, and the tolerance limits obtained by the SPC method are 98.98%, 97.74% and 98.62%, respectively. The tolerance limit calculated by SPC method is close to the optimal threshold of ROC, both of which could identify all errors of ±2 mm, while the universal tolerance limit can only partially identify them, indicating that the universal tolerance limit is not sensitive on some large errors. Therefore, considering the factors such as ease of use and accuracy, it is suggested to use the true composite measurement in clinical practice, and to formulate tolerance limits and action limits suitable for the actual process of the institution based on the SPC method. In conclusion, it is expected that the results of this study can provide some references for institutions to optimize the radiotherapy plan verification process, set appropriate pass rate limit, and promote the standardization of plan verification.

Long Pentraxin 3 as a New Biomarker for Diagnosis of Hip and Knee Periprosthetic Joint Infections.

Preoperative diagnosis of periprosthetic joint infections (PJIs) poses an unmet clinical challenge. The long pentraxin PTX3 is a component of the innate immune system involved in infection immunity. This study evaluated the potential of synovial and plasmatic PTX3 in the diagnosis of hip and knee PJIs. Consecutive total hip and knee arthroplasty (THA/TKA) revisions were prospectively included and classified as septic or aseptic according to the European Bone and Joint Infection Society (EBJIS) and Musculoskeletal Infection Society (MSIS) criteria. The concentration of PTX3 in plasma and synovial fluid samples was measured with ELISA. The AUC, threshold value, sensitivity, specificity, and positive and negative likelihood ratios were calculated using the ROC (receiver operating characteristic) curve method. The study population included 128 patients (94 THAs; 34 TKAs). The AUC of the synovial PTX3 based on EBJIS criteria was 0.85 (p < 0.0001), with a sensitivity of 81.13% and a specificity of 93.33%. The AUC based on MSIS criteria was 0.95 (p < 0.001), with a sensitivity of 91.43% and a specificity of 89.25%. Plasmatic PTX3 failed to discriminate infected from non-infected patients. Synovial PTX3 demonstrated an excellent diagnostic potential in hip and knee PJIs, with a very high specificity irrespective of the diagnostic criteria for PJI.

Radiomics Signature Using Manual Versus Automated Segmentation for Lymph Node Staging of Bladder Cancer

BackgroundBladder cancer (BC) treatment algorithms depend on accurate tumor staging. To date, computed tomography (CT) is recommended for assessment of lymph node (LN) metastatic spread in muscle-invasive and high-risk BC. However, the diagnostic efficacy of radiologist-evaluated CT imaging studies is limited. ObjectiveTo evaluate the performance of quantitative radiomics signatures for detection of LN metastases in BC. Design, setting, and participantsOf 1354 patients with BC who underwent radical cystectomy (RC) with lymphadenectomy who were screened, 391 with pathological nodal staging (pN0: n = 297; pN+: n = 94) were included and randomized into training (n = 274) and test (n = 117) cohorts. Pelvic LNs were segmented manually and automatically. A total of 1004 radiomics features were extracted from each LN and a machine learning model was trained to assess pN status using histopathology labels as the ground truth. Outcome measurements and statistical analysisRadiologist assessment was compared to radiomics-based analysis using manual and automated LN segmentations for detection of LN metastases in BC. Statistical analysis was performed using the receiver operating characteristics curve method and evaluated in terms of sensitivity, specificity, and area under the curve (AUC). Results and limitationsIn total, 1845 LNs were manually segmented. Automated segmentation correctly located 361/557 LNs in the test cohort. Manual and automatic masks achieved an AUC of 0.80 (95% confidence interval [CI] 0.69–0.91; p = 0.64) and 0.70 (95% CI: 0.58–0.82; p = 0.17), respectively, in the test cohort compared to radiologist assessment, with an AUC of 0.78 (95% CI 0.67–0.89). A combined model of a manually segmented radiomics signature and radiologist assessment reached an AUC of 0.81 (95% CI 0.71–0.92; p = 0.63). ConclusionsA radiomics signature allowed discrimination of nodal status with high diagnostic accuracy. The model based on manual LN segmentation outperformed the fully automated approach. Patient summaryFor patients with bladder cancer, evaluation of computed tomography (CT) scans before surgery using a computer-based method for image analysis, called radiomics, may help in standardizing and improving the accuracy of assessment of lymph nodes. This could be a valuable tool for optimizing treatment options.

Minimal important difference and patient acceptable symptom state for the Numerical Rating Scale (NRS) for pain and the Patient-Rated Wrist/Hand Evaluation (PRWHE) for patients with osteoarthritis at the base of thumb

BackgroundThe Numerical Rating Scale (NRS) and Patient-rated wrist/hand evaluation (PRWHE) are patient-reported outcomes frequently used for evaluating pain and function of the wrist and hand. The aim of this study was to determine thresholds for minimal important difference (MID) and patient acceptable symptom state (PASS) for NRS pain and PRWHE instruments in patients with base of thumb osteoarthritis.MethodsFifty-two patients with symptomatic base of thumb osteoarthritis wore a splint for six weeks before undergoing trapeziectomy. NRS pain (0 to 10) and PRWHE (0 to 100) were collected at the time of recruitment (baseline), after splint immobilization prior to surgery, and at 3, 6, 9 and 12 months after surgery. Four anchor-based methods were used to determine MID for NRS pain and PRWHE: the receiver operating characteristics (ROC) curve, the mean difference of change (MDC), the mean change (MC) and the predictive modelling methods. Two approaches were used to determine PASS for NRS pain and PRWHE: the 75th percentile and the ROC curve methods. The anchor question for MID was the change perceived by the patient compared with baseline; the anchor question for PASS was whether the patient would be satisfied if the condition were to stay similar. The correlation between the transition anchor at baseline and the outcome at all time points combined was calculated using the Spearman’s rho analysis.ResultsThe MID for NRS pain was 2.5 using the ROC curve method, 2.0 using the MDC method, 2.8 using the MC method, and 2.5 using the predictive modelling method. The corresponding MIDs for PRWHE were 22, 24, 10, and 20. The PASS values for NRS pain and PRWHE were 2.5 and 30 using the ROC curve method, and 2.0 and 22 using the 75th percentile method, respectively. The area under curve (AUC) analyses showed excellent discrimination for all measures.ConclusionWe found credible MID estimates for NRS and PRWHE (including its subscales), although the MID estimates varied depending on the method used. The estimates were 20-30% of the range of scores of the instruments. The cut-offs for MID and PASS showed good or excellent discrimination, lending support for their use in future studies.Trial registrationThis clinimetrics study was approved by the Helsinki University ethical review board (HUS1525/2017).

Evaluation of the prediction capability of AHP and F-AHP methods in flood susceptibility mapping of Ernakulam district (India)

Floods are one of the frequent natural hazards occurring in Kerala because of the remarkably high annual rate of rainfall. The objective of this study is to prepare the flood susceptibility maps of Ernakulam district by integrating remote sensing data, GIS, analytical hierarchy process (AHP), and fuzzy-analytical hierarchy process methods. Ernakulam is one of Kerala's most flood-prone districts. The development of this map can help to raise awareness about the risks of flooding. Factors such as slope angle, soil types (texture), land use/land cover, stream density, water ratio index, normalized difference built-up index, topographic wetness index, stream power index, aspect, and sediment transport index have been selected. The area of the final maps is grouped into five flood susceptible zones, ranging from very low to very high. The major reasons for flood occurrence in Ernakulam district are the combined effect of multiple factors such as excess silting, reduction of stream width due to anthropogenic activities, and changes in land cover and land-use pattern, lower slope, higher soil moisture content, lower stream capacity, and poor infiltration capacity of soils. The prepared map was validated using the receiver operating characteristic (ROC) curve method. The area under the ROC curve (AUC) values of 0.75 and 0.81 estimated by the ROC curve method for the AHP and F-AHP methods are considered acceptable and excellent, which confirms the prediction capability of the prepared maps. The very high susceptible zone constitutes around 19% of the district. This map is useful for land-use planners and policymakers to adopt strategies that will reduce the impact of flood hazards and damage in the future.

Multi attributive ideal-real comparative analysis (MAIRCA) method for evaluating flood susceptibility in a temperate Mediterranean climate

ABSTRACT This study is focused on flood susceptibility evaluation across the Golestan Province, Iran, using novel ensemble models generated by Multi Attributive Ideal-Real Comparative Analysis (MAIRCA) with frequency ratio (FR) and weight of evidence (WOE). As MAIRCA was employed in flood susceptibility assessment for the first time, an attempt has been made to evaluate its capability by comparing the ensemble with multilayer perceptron (MLP) neural network-based models. Ten flood conditioning factors (altitude, slope, aspect, plan curvature, distance from rivers, topographic wetness index, rainfall, soil type, geology, and land use) and 240 flood and non-flood locations were applied for modelling, of which 70% were selected for training and 30% for validation. The results of validation, performed by the receiver operating characteristics curve method, indicate that the highest predictive accuracy was obtained by MLP-WOE (0.926), followed by MLP-FR (0.912), MAIRCA-WOE (0.885), and MAIRCA-FR (0.859). High precision of the models implies their capability in flood risk prediction.

Central Sensitization Is an Important Factor in the Outcome of Patients Undergoing Total Knee Arthroplasty

Commentary It is well known that as many as 20% of patients are not satisfied after total knee arthroplasty (TKA), with such patients often reporting persistent pain or dysfunction1. Although there are many possible causes for dissatisfaction after TKA, it has become increasingly recognized in recent years that there are a multitude of factors outside of the surgical procedure that contribute to the overall success of TKA. An emerging area of research has sought to identify the role of central sensitization (CS) on the outcomes of TKA. Prior literature has demonstrated that patients with CS have worse quality of life, functional disability, and dissatisfaction than patients without CS after TKA despite achieving comparable clinical improvements after the surgical procedure2. This begs the question of whether CS impacts the calculation of the minimal clinically important difference (MCID) for outcome measures commonly utilized to evaluate patients who undergo TKA. Kim et al. attempted to answer this question by retrospectively evaluating 422 patients who underwent primary TKA. The authors preoperatively evaluated patients for CS utilizing a validated self-reported inventory for CS syndrome and obtained preoperative and 2-year postoperative Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores for all patients. The MCID was then calculated for patients with and without CS via both anchor-based and distribution methodology, utilizing both the change difference method and the receiver operating characteristic (ROC) curve method for anchor-based analyses. The authors should be commended for utilizing multiple methods for calculating the MCID, as this provides a much more inclusive evaluation of this important end point. One of the more interesting findings of this article was that 1 in 4 patients who underwent TKA in this series met the criteria for CS syndrome. Although this is not a new finding, it further highlights that CS is likely to be commonly encountered by orthopaedic providers caring for patients who undergo TKA. This study also demonstrated that the calculated MCIDs for patients with CS were significantly larger than those for patients without CS. Although the clinical importance of a 2 to 3-point difference in the MCID for the ROC curve and distribution-based methods of calculation has yet to be determined, the MCID was almost 9 points larger for patients with CS (23.4 points) compared with patients without CS (14.7 points) when calculated with the change difference method. This is a quite large and likely clinically important difference in MCID values. The trends from the data presented in the article are evident, and this study clearly demonstrated that patients with CS required larger improvements in clinical outcomes compared with patients without CS to reach a level that was considered a clinically meaningful improvement. This study further highlighted the complexity of caring for patients who underwent TKA and have CS by demonstrating that patients with CS were much less likely than patients without CS to even achieve an MCID after TKA. Depending upon the method of MCID determination, patients with CS achieved an MCID for the total WOMAC score at a rate of 13% to 15% less than patients without CS. Thus, not only do patients with CS require larger improvements in function to reach the MCID, but they are also significantly less likely to reach this outcome threshold. Although this article brings to light the importance of considering CS when evaluating MCIDs after TKA, it would have been helpful if the authors would have also evaluated the impact of CS on other defined clinical outcome thresholds such as the patient acceptable symptom state (PASS) and substantial clinical benefit (SCB), rather than focusing solely on the minimum improvement necessary to define success. This tiered approach to evaluating clinical outcomes scores (MCID, PASS, and SCB) provides a much more comprehensive view of patient outcomes3, and I suspect that the impact of CS would be even more magnified when analyzing higher definitions of clinical success. Nonetheless, Kim et al. demonstrate that CS is commonly encountered in patients who undergo TKA and is an important factor in their outcomes following the surgical procedure. CS should be considered when utilizing an MCID to compare interventions or outcomes between cohorts undergoing TKA going forward. Additionally, future research is now needed to identify ways in which we can intervene for patients with CS to bring their outcomes more closely in line with those of patients without CS.

Determination of Cutoff Values for the Screening of Osteosarcopenia in Obese Postmenopausal Women

Osteosarcopenic obesity (OSO) describes the concurrent presence of obesity, low bone mass, and low muscle mass in an individual. Currently, no established criteria exist to diagnose OSO. We hypothesized that obese individuals require different cut-points from standard cut-points to define low bone mass and low muscle mass due to their higher weight load. In this study, we determined cutoff values for the screening of osteosarcopenia (OS) in obese postmenopausal Malaysian women based on the measurements of quantitative ultrasound (QUS), bioelectrical impedance analysis (BIA), and functional performance test. Then, we compared the cutoff values derived by 3 different statistical modeling methods, (1) receiver operating characteristic (ROC) curve, (2) lowest quintile of the study population, and (3) 2 standard deviations (SD) below the mean value of a young reference group, and discussed the most suitable method to screen for the presence of OS in obese population. One hundred and forty-one (n = 141) postmenopausal Malaysian women participated in the study. Bone density was assessed using calcaneal quantitative ultrasound. Body composition was assessed using bioelectrical impedance analyzer. Handgrip strength was assessed using a handgrip dynamometer, and physical performance was assessed using a modified Short Physical Performance Battery test. ROC curve was determined to be the most suitable statistical modeling method to derive the cutoffs for the presence of OS in obese population. From the ROC curve method, the final model to estimate the probability of OS in obese postmenopausal women is comprised of five variables: handgrip strength (HGS, with area under the curve (AUC) = 0.698 and threshold ≤ 16.5 kg), skeletal muscle mass index (SMMI, AUC = 0.966 and threshold ≤ 8.2 kg/m2), fat-free mass index (FFMI, AUC = 0.946 and threshold ≤ 15.2 kg/m2), broadband ultrasonic attenuation (BUA, AUC = 0.987 and threshold ≤ 52.85 dB/MHz), and speed of sound (SOS, AUC = 0.991 and threshold ≤ 1492.15 m/s). Portable equipment may be used to screen for OS in obese women. Early identification of OS can help lower the risk of advanced functional impairment that can lead to physical disability in obese postmenopausal women.

Comparative study of statistical methods for clustered ROC data: nonparametric methods and multiple outputation methods

In clustered receiver operating characteristic (ROC) data each patient has several normal and abnormal observations. Within the same cluster, observations are naturally correlated. Several nonparametric methods have been proposed in the literature to handle clustered data structure, but their performances on simulated and real datasets have not been compared. Recently, a multiple outputation method has been considered for clustered data in areas other than diagnostic accuracy to account for within-cluster correlation. The multiple outputation method offers a resampling-based alternative for one sample clustered data with or without covariates, or for hypothesis testing in two sample clustered data. The method does not require a specific within-cluster correlation structure and yields a valid estimator while accounting for the within-cluster correlations. This paper contributes to the literature by introducing the multiple outputation method to the ROC setting, and empirically comparing the performance of these clustered ROC curve methods. The performance of these methods is also evaluated through two real examples.

Cutpoints for Muscle Mass and Strength Derived from Weakness or Mobility Impairment and Compared with Other Diagnostic Criteria in Community-Dwelling Elderly People.

We identified the strength cutpoints concerning mobility impairment, then identified the muscle mass cutpoints concerning weakness, and compared the results with other diagnostic criteria to develop the clinical diagnostic criteria associated with functional impairment. In 7583 elderly people, classification and regression tree (CART) and receiver operating characteristic curve (ROC) analyses were used for determining cutpoints for handgrip strength (HGS) and appendicular lean mass (ALM) indices associated with slowness or weakness. Logistic regressions were then used to quantify the strength of the association between muscle mass (or strength) categories and weakness (or slowness). The CART second cutpoints of muscle mass and strength indices were lower than those specified by the ROC method and were between those cutpoints determined by the 20th and Mean-2SD methods. After adjusting for covariates, the associations remained significant in handgrip strength categories defined by the CART and ROC cutpoints and HGS/BMI categories defined by the CART, ROC, and 20th cutpoints in men and women (P < 0.05), ALM, ALM/Ht2 categories defined by all four cutpoints (P < 0.05) and ALM/BMI categories defined by CART and ROC cutpoints in men (P < 0.05), and ALM and ALM/Ht2 categories defined by the CART cutpoints in women (P < 0.05). Our approaches resulted in a definition of weak strength as handgrip strength or HGS/BMI less than 26.55kg or 1.114 in men and less than 16.45kg or 0.697 in women and then defined ALM, ALM/Ht2, or ALM/BMI less than 18.92kg, 7.08kg/m2, or 0.795 in men and less than 15.04kg, 5.99kg/m2, or 0.517 in women as low lean mass.

Gonadotropin-Releasing Hormone Analogue Stimulation Test Versus Venous Sampling in Postmenopausal Hyperandrogenism.

Postmenopausal hyperandrogenism can be due to excessive androgen secretion from adrenal or ovarian virilizing tumors or nonneoplastic conditions. The etiology of postmenopausal hyperandrogenism can be difficult to discern because of limited accuracy of current diagnostic tests. This systematic review compares the diagnostic accuracy of the gonadotropin-releasing hormone (GnRH) analogue stimulation test against selective ovarian and adrenal vein sampling of androgens in distinguishing neoplastic from nonneoplastic causes of postmenopausal hyperandrogenism. Diagnostic test accuracy studies on these index tests in postmenopausal women were selected based on preestablished criteria. The true positive, false positive, false negative, and true negative values were extracted and meta-analysis was conducted using the hierarchical summary receiver operator characteristics curve method. The summary sensitivity of the GnRH analogue stimulation test is 10% (95% confidence interval [CI], 1.1%-46.7%) and that for selective venous sampling is 100% (95% CI, 0%-100%). Both tests have 100% specificity. There is limited evidence for the use of either test in identifying virilizing tumors in postmenopausal hyperandrogenism.

Minimal important difference and patient acceptable symptom state for PFDI-20 and POPDI-6 in POP surgery

Introduction and hypothesisPatient-reported outcome measures are fundamental tools when assessing effectiveness of treatments. The challenge lies in the interpretation: which magnitude of change in score is meaningful for the patients? The minimal important difference (MID) is defined as the smallest difference in score that patients perceive as important. The Patient Acceptable Symptom State (PASS) represents the value of score beyond which patients consider themselves well. We aimed to determine the MID and PASS for Pelvic Floor Distress Inventory-20 (PFDI-20) and Pelvic Organ Prolapse Distress Inventory-6 (POPDI-6) in pelvic organ prolapse (POP) surgery.MethodsWe used data from 2704 POP surgeries from a prospective, population-based cohort. MID was determined with three anchor-based and one distribution-based method. PASS was defined using two different methods. Medians of the estimates were identified.ResultsThe MID estimates with (1) mean change, (2) receiver-operating characteristic (ROC) curve, (3) 75th percentile, and (4) distribution-based method varied between 22.9–25.0 (median 24.2) points for PFDI-20 and 9.0–12.5 (median 11.3) for POPDI-6. The PASS cutoffs with (1) 75th percentile and (2) ROC curve method varied between 57.7–62.5 (median 60.0) for PFDI-20 and 16.7–17.7 (median 17.2) for POPDI-6.ConclusionA mean difference of 24 points in the PFDI-20 or 11 points in the POPDI-6 can be used as a clinically relevant difference between groups. Postoperative scores ≤ 60 for PFDI-20 and ≤ 17 for POPDI-6 signify acceptable symptom state.

High Serum Levels of Serum 100 Beta Protein, Neuron-specific Enolase, Tau, Active Caspase-3, M30 and M65 in Children with Autism Spectrum Disorders.

ObjectiveThe purpose of this study was therefore to investigate whether neuronal, axonal, and glial cell markers (Neuron-specific enolase [NSE], tau, serum 100 beta protein [S100B], respectively) and apoptosis markers (active caspase 3, M30, M65) and whether these parameters can be used as diagnostic biomarkers in autism spectrum disorders (ASD).MethodsThis study measured the serum S100B, NSE, tau, active caspase 3, M30, and M65 levels in 43 patients with ASD (aged 3−12 years) and in 41 age- and sex-matched healthy controls. ASD severity was rated using the Childhood Autism Rating Scale. The serum levels were determined in the biochemistry laboratory using the ELISA technique. The receiver operator characteristics curve method was employed to evaluate the accuracy of the parameters in diagnosing ASD.ResultsSerum S100B, tau, NSE, active caspase-3, M30, and M65 levels were significantly higher in the patient group than in the control group (p < 0.001, p = 0.002, p = 0.002, p = 0.005, p < 0.001, and p = 0.004, respectively). The cut-off value of S100B was 48.085 pg/ml (sensitivity: 74.4%, specificity: 80.5%, areas under the curve: 0.879, p < 0.001).ConclusionApoptosis increased in children with ASD, and neuronal, axonal, and glial cell injury was observed. In addition, S100B may be an important diagnostic biomarker in patients with ASD. Apoptosis, and neuronal, axonal and astrocyte pathologies may play a significant role in the pathogenesis of ASD, and further studies are now required to confirm this.

Influence of time between surgery and postoperative radiation therapy and total treatment time in locoregional control of patients with head and neck cancer: a single center experience

OBJECTIVE:This study aimed to evaluate the effect of the delay to initiate postoperative radiation therapy (RT) on locoregional control to head and neck squamous cell carcinoma patients.METHODS:Retrospective cohort study that included patients submitted to surgery followed by adjuvant RT (with/without chemotherapy). The time interval between surgery and RT was dichotomized by the receiver operating characteristics curve method at 92 days. Other possible sources of heterogeneity with potential impact on locoregional control were explored by regressive analysis.RESULTS:A total of 168 patients were evaluated. The median time for locoregional recurrence (LRR) was 29.7 months. The relapse-free survival rates were 66.4% and 75.4% for patients who initiated RT more than and within 92 postoperative days (p=0.377), respectively. Doses lower than 60Gy were associated with worse rates of locoregional control (HR=6.523; 95%CI:2.266-18.777, p=0.001). Patients whose total treatment time (TTT) was longer than 150 days had LRR rate of 41.8%; no patient with TTT inferior to 150 days had relapses (p=0.001).CONCLUSIONS:The interval between surgery and RT did not show influence on locoregional control rates. However, doses <60Gy and the total treatment time >150 days were associated with lower locoregional control rates.

Association of high sensitive C-reactive protein with coronary heart disease: a Mendelian randomization study

ObjectivesWhether high sensitivity C-reactive protein (hs-CRP) has a causal effect on coronary heart disease (CHD) is unclear. This study investigated the causal effect of hs-CRP on CHD risk using Mendelian Randomization (MR) analysis.MethodsA total of 3802 subjects were recruited in the follow-up study. Linear regression model was used to evaluate the relationship between CRP polymorphisms and hs-CRP. Survival receiver operator characteristic curve method was used to explore the cut-off of hs-CRP on CHD incidence. Cox regression model was applied to detect the association of hs-CRP with CHD by calculating the hazard ratio (HR) and 95% confidence interval (CI). Rs1205 and rs876537 in CRP were selected as instrumental variables in MR analysis.ResultsDuring a median follow-up time of 5.01 years, 98 CHD incidence was identified (47.03/104 person-years). Hs-CRP was significantly increased among rs1205 and rs876537 genotypes with r values of 0.064 and 0.066, respectively. Hs-CRP 1.08 mg/L was identified as the cut-off value with a maximum value of sensitivity and specificity on prediction of CHD. Participants with ≥1.08 mg/L of hs-CRP has a higher risk of CHD incidence than that of participants with < 1.08 mg/L, the adjusted HR (95% CI) was 1.69 (1.11–2.60) with a P value of 0.016. No significant casual association was observed between hs-CRP and CHD with a P value of 0.777.ConclusionsThe association between hs-CRP and CHD is unlikely to be causal, hs-CRP might be a predictor for incidence of CHD in general population.

Spatial characterization and distribution modelling of Ensete ventricosum (wild and cultivated) in Ethiopia

Enset (Ensete ventricosum) feeds around 20 million people in Ethiopia and is arguably the most important crop for food security and rural livelihoods in the country. Therefore, it is significant to know its spatial characterization and distribution in the country. We use spatial overlay analysis and the maximum entropy (MaxEnt) model for characterizing and modelling, respectively. Inputs for the model include 26 environmental variables—19 bioclimatic and seven biophysical—in addition to the geospatial location of enset field data. The model result was validated using Receiver Operating Characteristic curve method and the area under the curve (AUC) with 0.842 for cultivated enset and 0.760 (wild enset). The highest prediction (>0.5) of both varieties occurred in the southwestern, south-central and north-eastern parts of Ethiopia—17,293.67 km2 (cultivated) and 40,402 km2 (wild) area. The presence of both enset is probabilistically higher at the tropic-cool/sub-humid Agroecological Zones.