Robust Evaluation Research Articles

Clinical Introduction of a Volumetric-Based Skin-Sparing Planning Technique for Breast and Chest Wall Patients Treated with PBS Proton Therapy: A Review on Feasibility and Impact on Target Coverage and Robustness

<h3>Purpose/Objective(s)</h3> Acute and late skin toxicities are common and bothersome for patients undergoing breast or chest wall (CW) irradiation. There is clinical concern that acute skin toxicity may be higher with proton radiation compared with photon radiation because of differences in their energy deposition properties. Limited literature exists on dosimetric strategies to mitigate skin toxicity for patients treated with intensity modulated proton therapy (IMPT). We characterize the outcomes of a volumetric-based skin-sparing planning technique for IMPT treated breast and CW patients. <h3>Materials/Methods</h3> Prior to a consensus strategy, skin dose evaluation in IMPT Breast and CW plans was heterogeneous across physicians and dosimetrists. Uniformly, clinical targets were delineated by cropping off the skin surface by 3 mm for CW and 5 mm for intact breast; however, further skin-sparing strategies were previously restricted to limiting hot spots. In January 2022 our center established a consensus strategy to mitigate skin toxicity that added volumetric-based skin sparing objectives in addition to hot-spot evaluation. To start, a consistent skin evaluation structure (skin-eval) was defined within 5 mm of the CTV and included a rind of 3 mm for CW and 5 mm for intact breast, bound by the patient's surface. IMPT planning incorporated an objective to limit the volume of skin-eval receiving 95% of the prescription dose or more (V95%) to ideally less than 40% (goal of less than 50%) of the volume of the skin-eval structure, Skin-eval V95%Rx<40% (goal 50%), while still prioritizing CTV coverage and robustness. We report the early dosimetric experience with this volumetric-based objective. Boost plans were not included in this study. <h3>Results</h3> Ten patients underwent IMPT planning incorporating the described skin-eval objective, including 6 with intact breasts and 4 CW. Median skin-eval V95%Rx=36.4% (range, 20.3-49.9%). Median skin-eval max dose to 0.03cc, 0.3cc, and 1cc was 102.8% (range 101.4%-103.9%), 101.8% (range 100.5%-102.9%), and 101.1% (range 99.5%-102.3%), respectively. Median CTV V97.5%Rx was 98.2% (range, 94.9%-99.5%). 9 of the 10 plans passed all robust evaluation scenarios at V95%Rx receiving at least 95% (median 97.7%, range 96.1%-98.5%). Only 1 plan did not meet CTV goal coverage in all robust scenarios, failing only 1 scenario. This plan uniquely incorporated a couch kick to address atypical setup challenges. <h3>Conclusion</h3> A volumetric-based objective of skin-eval V95%Rx<50% appears consistently achievable while still maintaining excellent nominal and robustly optimized target coverage in patients receiving IMPT to the breast and CW. Future studies with both retrospective and prospective designs can validate dose constraints that may be more relevant for toxicity mitigation with IMPT.

How can policymakers incorporate uncertainty (as modelled through foresight) into policy evaluation?

Abstract Background Using a robust policy evaluation framework is critical to inform policymakers about the relevance of a policy in a structured way, clarifying its implications and providing arguments and criteria to compare competing options and decide which policy should be prioritised. In the current context of complex societal, environmental and public health challenges, decision-makers demand tools that allow a comparative assessment of the effectiveness of multisectoral policies and their comparison under different scenarios. Methods This presentation will detail a socio-technical “desirability-doability” framework (2xD). In 2xD, an additive value model is constructed to measure the desirability of public policies, and their doability is appraised under two contrasted scenarios. For these purposes, the MACBETH approach is used in developing the three group modelling phases of 2xD: (I) Structuring facilitated workshops, (II) Evaluation decision conference, and (III) Desirability-doability decision conference. Tailor-made interactive protocols or questioning procedures are used to elicit group judgements, based on which objective-specific value scores are assigned to the policies, and the objectives are weighted (in II). Finally, we elicit doability scores for the policies under each scenario (in III). Results We present how the 2xD framework was applied in Lisbon's urban health policymaking setting and discuss case insights regarding the role of scenario analysis in the appraisal and selection of policies. Conclusions The 2xD framework advances knowledge on how to assist policymakers in evaluating and selecting policies and contributes to the literature on overall policy formulation and evaluation. Specifically, the framework enabled a group of policymakers to balance (multicriteria) desirability versus doability of policies under the light of two contrasting scenarios, therefore incorporating uncertainty in their decision-making process.

International Research and Philanthropy

International Research and Philanthropy

Wellbore Integrity After a Blowout: Stress Evolution Within the Casing-Cement Sheath-Rock Formation System

A robust evaluation of the casing-cement sheath-rock formation system is foundational in ensuring wellbore integrity throughout a well's lifecycle. Analytical theory for a multi-layer, thick-walled cylinder is used to evaluate the aggregate stress distributions within the casing-cement sheath-rock formation system, honoring boundary conditions of radial stress and displacement continuity. These stress distributions are adjusted for scenarios that a well may experience in its lifetime. Each layer of the casing-cement sheath-rock formation system is evaluated separately against various possible mechanical failure mechanisms, all of which can compromise wellbore integrity.A blowout scenario after a mismanaged loss-of-well-control situation induces high stress loads in the casing-cement sheath-rock formation system, with the wellbore pressure rapidly decreasing during post-blowout discharge, followed by a rapid increase following successful well capping. Casing and cement sheath failures can expose the surrounding rock formation to the pressurized fluid inside the wellbore, risking hydrocarbons broaching the surface, or the seafloor. The aggregate stress distribution model is applied on a case study performed using parameters from the MC 252–1 “Macondo Well” blowout from April 20, 2010 in the deepwater Gulf of Mexico. The stress evolution suggests stability against mechanical failure mechanisms within the casing (collapse/burst and tensile/compressive), cement-sheath (inner or outer debonding, and shear cracking), and rock-formation layers (longitudinal or transverse tensile fracture initiation, along with shear-slippage along pre-existing faults in the near-well vicinity), throughout the blowout aftermath. Nevertheless, tendencies towards radial cracking and disking (tensile) failures were indicated for the cement-sheath layer as the system reaches radial stress and displacement continuity, after cement setting.

Robust evaluation of coercivity in exchange biased films

Exact measurement of coercivity in thin magnetic films is of high importance for sensor industry. In exchange biased multilayers, incorporated in magnetoresistive sensors, coercivity of the coupled ferromagnetic layers strongly depends on the orientation of easy axis in respect to field. We show, that in NiFe/IrMn films with high HB/HC ratio, field misalignment ∼2° results in ∼50 % change in coercivity. Inhomogeneous magnetic field in a film deposition device as well as manual installation of the sample in magnetometers result in large error in coercivity. In this work, we propose measurement and analysis protocol to avoid this uncertainty and to evaluate easy axis coercivity values from tilted angle measurements in a wide temperature range using just one reference value of easy axis coercive field, measured at room temperature.

Updated database, semi-empirical and theoretical calculation of K[formula omitted]/K[formula omitted] intensity ratios for elements ranging from [formula omitted]Na to [formula omitted]Cm

A collection of 1118 experimental Kβ/Kα X-ray intensity ratios induced by photon impact was published recently by Daoudi et al. [1]. There were errors in some tabulated values. These errors are corrected here, as well as a substantial update including all experimental Kβ/Kα X-ray intensity ratios obtained from various sources for elements ranging from 11Na to 96Cm, when bombarded with photons, protons, deuterons, alpha particles, and electrons. More than 2100 experimental Kβ/Kα X-ray intensity ratios published between 1969 and 2020 (196 papers) are summarized in a table. A thorough analysis of these data is provided, as well as a table with weighted average intensity ratio values Kβ/KαW. Subsequently new semi-empirical average Kβ/Kα X-ray intensity ratios for elements in the range 11 ≤ Z ≤ 96 have been determined using an interpolation procedure which involves the analytical function Kβ/KαW/1−Kβ/KαW1/4 as a function of the atomic number Z and then fitting the experimental to weighted average ratios S=Kβ/KαEXP/Kβ/KαW. Furthermore, new theoretical calculations based on the Multiconfiguration Dirac–Fock Method have been performed for some elements and are presented in this work. The compilation provides a snapshot of the current status of Kβ/Kα X-ray atomic data. The need for additional quality measurements with detailed uncertainty assessments is clear. Considerable expert judgment will be needed in the future to shift through the database to create a statistically robust evaluation.

Machine-learning model predicting postoperative delirium in older patients using intraoperative frontal electroencephalographic signatures.

ObjectiveIn older patients receiving general anesthesia, postoperative delirium (POD) is the most frequent form of cerebral dysfunction. Early identification of patients at higher risk to develop POD could provide the opportunity to adapt intraoperative and postoperative therapy. We, therefore, propose a machine learning approach to predict the risk of POD in elderly patients, using routine intraoperative electroencephalography (EEG) and clinical data that are readily available in the operating room.MethodsWe conducted a retrospective analysis of the data of a single-center study at the Charité-Universitätsmedizin Berlin, Department of Anesthesiology [ISRCTN 36437985], including 1,277 patients, older than 60 years with planned surgery and general anesthesia. To deal with the class imbalance, we used balanced ensemble methods, specifically Bagging and Random Forests and as a performance measure, the area under the ROC curve (AUC-ROC). We trained our models including basic clinical parameters and intraoperative EEG features in particular classical spectral and burst suppression signatures as well as multi-band covariance matrices, which were classified, taking advantage of the geometry of a Riemannian manifold. The models were validated with 10 repeats of a 10-fold cross-validation.ResultsIncluding EEG data in the classification resulted in a robust and reliable risk evaluation for POD. The clinical parameters alone achieved an AUC-ROC score of 0.75. Including EEG signatures improved the classification when the patients were grouped by anesthetic agents and evaluated separately for each group. The spectral features alone showed an AUC-ROC score of 0.66; the covariance features showed an AUC-ROC score of 0.68. The AUC-ROC scores of EEG features relative to patient data differed by anesthetic group. The best performance was reached, combining both the EEG features and the clinical parameters. Overall, the AUC-ROC score was 0.77, for patients receiving Propofol it was 0.78, for those receiving Sevoflurane it was 0.8 and for those receiving Desflurane 0.73. Applying the trained prediction model to an independent data set of a different clinical study confirmed these results for the combined classification, while the classifier on clinical parameters alone did not generalize.ConclusionA machine learning approach combining intraoperative frontal EEG signatures with clinical parameters could be an easily applicable tool to early identify patients at risk to develop POD.

A novel abnormality annotation database for COVID-19 affected frontal lung X-rays.

Consistent clinical observations of characteristic findings of COVID-19 pneumonia on chest X-rays have attracted the research community to strive to provide a fast and reliable method for screening suspected patients. Several machine learning algorithms have been proposed to find the abnormalities in the lungs using chest X-rays specific to COVID-19 pneumonia and distinguish them from other etiologies of pneumonia. However, despite the enormous magnitude of the pandemic, there are very few instances of public databases of COVID-19 pneumonia, and to the best of our knowledge, there is no database with annotation of abnormalities on the chest X-rays of COVID-19 affected patients. Annotated databases of X-rays can be of significant value in the design and development of algorithms for disease prediction. Further, explainability analysis for the performance of existing or new deep learning algorithms will be enhanced significantly with access to ground-truth abnormality annotations. The proposed COVID Abnormality Annotation for X-Rays (CAAXR) database is built upon the BIMCV-COVID19+ database which is a large-scale dataset containing COVID-19+ chest X-rays. The primary contribution of this study is the annotation of the abnormalities in over 1700 frontal chest X-rays. Further, we define protocols for semantic segmentation as well as classification for robust evaluation of algorithms. We provide benchmark results on the defined protocols using popular deep learning models such as DenseNet, ResNet, MobileNet, and VGG for classification, and UNet, SegNet, and Mask-RCNN for semantic segmentation. The classwise accuracy, sensitivity, and AUC-ROC scores are reported for the classification models, and the IoU and DICE scores are reported for the segmentation models.

Effects of global climate change on the hydrological cycle and crop growth under heavily irrigated management – A comparison between CMIP5 and CMIP6

Quantifying the impact of global climate change on the water cycle and crop production is essential for water resource management and agricultural production planning. This study compared multiple GCM (General Circulation Model) projections of CMIP5 (Coupled Model Intercomparison Project 5) and CMIP6 and further used the selected GCMs to quantify the impact of future climate change on hydrology and crop production under intensive irrigation management in the North China Plain using an improved SWAT model (SWAT-MAD). Taylor’s skill score was used firstly to screen out six groups of GCMs with better simulation performance from 10 pairs of homologous GCMs of CMIP5 and CMIP6. The selected GCMs of CMIP5 and CMIP6 were further used to drive SWAT-MAD for a robust evaluation of climate change impacts. Results showed that during winter wheat growing season, average actual evapotranspiration (ETa) increased by 3%, 3%, 4%, and 5%, respectively, under 2041–2070 RCP4.5, 2041–2070 SSP2-4.5, 2041–2100 RCP4.5, and 2071–2100 SSP2-4.5 scenarios, compared to the historical period (1971–2000). During summer maize growing season, those changes in ETa were 4%, 2%, −0.2%, and −3%. Predicted future precipitation, air temperatures, and surface runoff could increase, while irrigation could decrease as precipitation increased. The dynamic patterns of leaf area index of winter wheat and summer maize indicated that there is a tendency for early emergence and maturity of both crops in the future, and the daily total biomass elevated with a corresponding increase in final yields. The highest increases in yields of winter wheat and summer maize were 18.9% and 16.7%. The findings not only contribute to enhancing the confidence of future projections using CMIP6 but also facilitate our understanding of the relative uncertainty of GCMs. This study provides technical and data support for pre-selections of GCMs and decision making in best management practices for groundwater conservation and agricultural production.

Pre-treatment analysis of non-rigid variations can assist robust intensity-modulated proton therapy plan selection for head and neck patients.

To incorporate small non-rigid variations of head and neck patients into the robust evaluation of intensity-modulated proton therapy (IMPT) for the selection of robust treatmentplans. A cohort of 20 nasopharynx cancer patients with weekly kilovoltage CT (kVCT) and 15 oropharynx cancer patients with weekly cone-beam CT (CBCT) were retrospectively included. Anatomical variations between week 0/week 1 of treatment were acquired using deformable image registration (DIR) for all 35 patients and then applied to the planning CT of four patients who have kVCT scanned each week to simulate potential small non-rigid variations (sNRVs). The robust evaluations were conducted on IMPT plans with: (1) different number of beam fields from 3-field to 5-field; (2) different beam angles. The robust evaluation before treatment, including the sNRVs and setup uncertainty, referred to as sNRV+R evaluation was compared with the conventional evaluation (without sNRVs) in terms of robustness consistency with the gold standard evaluation based on weeklyCT. Among four patients (490 scenarios), we observed a maximum difference in the sNRV+R evaluation to the nominal dose of: 9.37% dose degradation on D95 of clinical target volumes (CTVs), increase in mean dose (D ) of parotid 11.87Gy, increase in max dose (D ) of brainstem 20.82Gy. In contrast, in conventional evaluation, we observed a maximum difference to the nominal dose of: 7.58% dose degradation on D95 of the CTVs, increase in parotid D by 4.88Gy, increase in brainstem D by 13.5 Gy. In the measurement of the robustness ranking consistency with the gold standard evaluation, the sNRV+R evaluation was better or equal to the conventional evaluation in 77% of cases, particularly, better on spinal cord, parotid glands, and low-riskCTV. This study demonstrated the additional dose discrepancy that sNRVs can make. The inclusion of sNRVs can be beneficial to robust evaluation, providing information on clinical uncertainties additional to the conventional rigid isocentershift.

Study of Ultrasonic-Guided Wave Interaction With Core Crush Damage for Nondestructive Evaluation of a Honeycomb Composite Sandwich Panel

AbstractHoneycomb composite sandwich structures are extensively used for the manufacturing of many different components of aerospace, automobiles, wind turbine blades, and marine ship hull structures. Despite its widespread use and advantages, the honeycomb core is frequently damaged during production and operation, even if the damage is not visible on the face-sheet. In this study, an ultrasonic-guided wave (GW) propagation technique is utilized for robust and reliable nondestructive evaluation of a honeycomb composite sandwich panel (HCSP) in the presence of core crush damage. A 2D semi-analytical model was developed to understand the dispersion characteristics in the HCSP and to identify various modes of GW propagation in the signals. Extensive numerical simulations are carried out using abaqus to study the guided wave interaction with core crush damage. For this purpose, two numerical models were considered (a realistic model with both crushed core and cavity, and a simplified model that only comprises the cavity) and experimentally validated using a contact-type transducer. The presence of core crush damage in an HCSP increases the amplitude and group velocity of the primary antisymmetric mode, and this characteristic has been used for localization of the core crush region in the HCSP. Finally, a damage detection algorithm using signal difference coefficient is presented for successful localization of the core crush region within a square monitoring area. Unlike other studies reported in the literature, we demonstrate the utility of the simplified numerical model for studying GW interactions with core crush defect and experimentally validate the nondestructive evaluation (NDE) technique to localize core crush defect on an HCSP.

Machine Learning Techniques for Device-Free Localization Using Low-Resolution Thermopiles

Indoor device-free localization (DFL) has many uses, including aged care, location-based services, ambient-assisted living, and fire safety management. In recent publications, thermopile sensors (very low-resolution infrared cameras) have been shown as being able to localize individuals while preserving their privacy. This article reports the performance evaluation of a large number of supervised machine learning techniques for the localization of a target using a ceiling-mounted thermopile. The algorithms were trained and validated using a large data set constructed from an individual walking arbitrary paths with the accurate ground truth provided by a virtual reality system. For robust performance evaluation, the algorithms were tested with data sets collected on a different day with several other subjects. A 2-D convolutional neural network exploiting spatial correlation and several recurrent neural network structures exploiting temporal correlation among the captured data provided the most accurate localization performance. Several data sets, constructed from the thermopile’s readings for four individual targets, were made available online for other researchers to use.

RosettaStone: Connecting the Past, Present, and Future of Physical Design Research

<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Editor’s notes:</i> In this article, the authors introduce RosettaStone, an open and extensible foundation, which leverages a standard physical design data model (LEF/DEF 5.8) and open-source database implementation (OpenDB) to effectively connect the academic physical design field’s past, present, and future. The authors also show how RosettaStone enables integration of closed-source research tools and nonstandard data formats for robust evaluation with modern technologies and testcases. Following the release of the IEEE CEDA DATC Robust Design Flow, RosettaStone is expected to contribute to a standard backplane for integration and testing of new methods in academic research on physical design and its translation into industry practice. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">—Yiran Chen, Duke University</i>

Overdiagnosis and overuse of diagnostic and screening tests in low-income and middle-income countries: a scoping review

ObjectiveOverdiagnosis and overuse of healthcare services harm individuals, take resources that could be used to address underuse, and threaten the sustainability of health systems. These problems are attracting increasing attention...

Bedside monitoring of lung perfusion by electrical impedance tomography

As a non-invasive and radiation-free bedside imaging method, electrical impedance tomography (EIT) can perform real-time regional pulmonary ventilation evaluation and pulmonary blood flow monitoring for patients, thus realizing bedside ventilation/perfusion matching visualization, effectively guiding the pathophysiological mechanism of hypoxemia, and providing a new method for the study of pulmonary blood flow. EIT has also played a unique and irreplaceable role in COVID-19 research and treatment. At the same time, as functional imaging, the operation details and image reconstruction algorithm of this technology still need to be further optimized by more researches to provide a more robust evaluation in clinical application. In this paper, EIT pulmonary blood flow monitoring methods, operation and implementation of monitoring indicators, application and related research progress will be described.

Evaluation and comparison of unsupervised methods for the extraction of spatial patterns from mass spectrometry imaging data (MSI)

For the extraction of spatially important regions from mass spectrometry imaging (MSI) data, different clustering methods have been proposed. These clustering methods are based on certain assumptions and use different criteria to assign pixels into different classes. For high-dimensional MSI data, the curse of dimensionality also limits the performance of clustering methods which are usually overcome by pre-processing the data using dimension reduction techniques. In summary, the extraction of spatial patterns from MSI data can be done using different unsupervised methods, but the robust evaluation of clustering results is what is still missing. In this study, we have performed multiple simulations on synthetic and real MSI data to validate the performance of unsupervised methods. The synthetic data were simulated mimicking important spatial and statistical properties of real MSI data. Our simulation results confirmed that K-means clustering with correlation distance and Gaussian Mixture Modeling clustering methods give optimal performance in most of the scenarios. The clustering methods give efficient results together with dimension reduction techniques. From all the dimension techniques considered here, the best results were obtained with the minimum noise fraction (MNF) transform. The results were confirmed on both synthetic and real MSI data. However, for successful implementation of MNF transform the MSI data requires to be of limited dimensions.

Evaluating Attribution in Dialogue Systems: The BEGIN Benchmark

Abstract Knowledge-grounded dialogue systems powered by large language models often generate responses that, while fluent, are not attributable to a relevant source of information. Progress towards models that do not exhibit this issue requires evaluation metrics that can quantify its prevalence. To this end, we introduce the Benchmark for Evaluation of Grounded INteraction (Begin), comprising 12k dialogue turns generated by neural dialogue systems trained on three knowledge-grounded dialogue corpora. We collect human annotations assessing the extent to which the models’ responses can be attributed to the given background information. We then use Begin to analyze eight evaluation metrics. We find that these metrics rely on spurious correlations, do not reliably distinguish attributable abstractive responses from unattributable ones, and perform substantially worse when the knowledge source is longer. Our findings underscore the need for more sophisticated and robust evaluation metrics for knowledge-grounded dialogue. We make Begin publicly available at https://github.com/google/BEGIN-dataset.

Evaluating chemical hazard on a global scale

Approaches to address the growing burden of chemical related diseases globally has not kept pace with this increasing human health threat. Policy and regulatory structures for addressing chemical exposures vary globally, but all incorporate elements of scientific evidence collection and evaluation. More transparent, consistent methods for hazard identification including systematic reviews have been implemented and adopted by various countries and global agencies. This talk will discuss opportunities to standardize and improve hazard identification that can be shared between high resource and low resource countries. More consistent approaches that incorporate contemporary scientific principles for robust data evaluation will improve the basis for health-related decision making.

Traffic-related air pollution and birth weight: the roles of noise, placental function, green space, physical activity, and socioeconomic status (FRONTIER)

Background and aim: FRONTIER aims to study the impact of maternal exposure to traffic-related air pollution (TRAP) on fetal growth while disentangling the effects of noise, elucidating the role of placental function as an underlying mechanism; and exploring the potential of greenspace to mitigate it. Methods: FRONTIER has established a new pregnancy cohort of 1,086 pregnant women in Barcelona (BiSC), Spain. Fetal growth was characterized by anthropometric measures at birth together with ultrasound-based measures of fetal development. We are integrating objective data on time-activity patterns with a hybrid modeling framework combining dispersion and land use regression models and campaigns of personal and home-outdoor air pollution monitoring to estimate maternal exposure level and inhaled dose of NO2, PM2.5, and PM2.5 light absorption and contents at the main microenvironments (home, workplace, and commuting routes). We are assessing maternal exposure to noise by integrating measurements at participants’ home-outdoor using noise monitors together with modeled microenvironmental noise levels and data on noise sensitivity, annoyance, and protections against noise. We will develop single- and multi-pollutant models to evaluate the impact of air pollution exposure and inhaled dose on fetal growth and the mediatory role of placental function. Results: BiSC mothers had an average age of 34.8 years with 31.1% having high school education and less and 73.6% having European origin. The mean (SD) of the first and third trimester measured personal NO2 levels were respectively 31.9(11.7) and 31.0(12.1)&amp;#x3bc;g/m3 for the pre-pandemic period and 26.6(10.9) and 25.1(12.1)&amp;#x3bc;g/m3 for the pandemic period. The mean (SD) of measured home-outdoor noise levels (LAeq,24h) was 61.5 (6.7)dB(A). 49.2% of newborns were girl and the mean (SD) of birthweight was 3276.8 (508.2)g The analyses are still ongoing. Conclusions: FRONTIER will provide a robust and comprehensive evaluation of the impact of TRAP exposure on fetal growth using high-resolution exposure data.

Is conductivity measurement or inductively coupled plasma-atomic emission spectrometry reliable to define rejection of different ions?

Rejection of single salts or ions is a basic and crucial characteristic of nanofiltration (NF) membranes. The simple and most pursued method to quantify the salt concentration has been via conductivity measurement. Pitfalls exist when ions hydrolysis or feed water contains monovalent ions. This could be explained in two possible scenarios: (1) easily hydrolyzed single salts form low charged ions and reduce feed pH, resulting in increased permeate conductivity and low nominal rejection; (2) for membranes with high multivalent ion rejections (>99%) or the concentration of target ions in feed is low, conductivity measurement results in low rejection due to the passage of monovalent ions if deionized water is used for the feed solution. A correction formula by subtracting the concentration of monovalent ions in water to obtain an accurate rejection value is proposed. This work provides an accurate, simple and robust evaluation of rejection for NF membranes, which promotes fair comparison of performance in literature, reliable analysis of separation mechanisms as well as a precise determination of product purity.