Current Metrics Research Articles

Mobile Broadband Performance Evaluation: Analysis of National Reports

Five decades have passed since the first bit was transmitted over the internet. Although the internet has improved our lives and led to the digital economy, currently only 51% of the world’s population have access to it. Currently, consumers mostly access the internet via mobile broadband, 2G, 3G, and 4G services. Regulatory bodies such as the Federal Communications Commission (FCC) of the US are responsible for ensuring that consumers receive an adequate service from Mobile Network Operators (MNOs). Usually, regulators evaluate the performance of each MNO in terms of service quality yearly and publish a report. To evaluate performance, metrics such as coverage, download/upload speed, and the number of subscribers can be used. However, the evaluation process and the metrics used by each regulatory body are inconsistent, and this makes it hard to determine which nations are providing adequate services to their citizens. Furthermore, it is not clear as to which performance evaluation is the right path. In this case study, we analyzed the reports released from eight nations (United States of America, United Kingdom, France, South Korea, Japan, Singapore, and Australia) as of the year 2020. We then point out the advantages and the drawbacks of the current evaluation process and metrics. Furthermore, a discussion on why the current methods are not sufficient to evaluate 5G services is presented. Our findings indicate that there is a great need for a unified metric and that this process becomes more complex with the rollout of 5G.

Towards standards in the analysis of wind turbines operating in cold climate. Part B: Methodology for evaluating wind turbine alternative operational strategies

Alternative operational strategies are now being used to cope with the challenges of icing on wind turbines. However, there are currently no generally accepted standards for the performance analysis of wind turbines using such alternative operational strategies. Hence, this is an important research topic that needs to be addressed. The aforementioned alternative operational strategies include ice protection systems (installation or modification) and alternative turbine control strategies. The interest surrounding the performance analysis of such strategies is rising since, as the demand for ice protection systems from wind turbine original equipment manufacturer and retrofits providers is increasing worldwide, operators are requesting warranties surrounding their performance. Due to their novelty, operators often prefer to test alternative operational strategies on a few turbines before deploying them on every turbines of a wind farm. In those situations, a standard and robust method is needed to evaluate the strategy before a full-scale extension. This paper proposes a methodology based on the experience gathered on five undisclosed wind farms where different operational strategies were implanted. Through case studies, the current metrics used to assess the performance of alternative operational strategies are reviewed and two new metrics that address the main issues with the current indicators are proposed. These new metrics can estimate the energy reclaimed through the use of alternative operational strategies. This paper also presents a methodology for selecting the best turbines that will be used as a reference to evaluate the performance of the turbines with the alternative operational strategies. Standardized metrics with the right turbines used as a reference will help to provide a clearer overview of the performance of wind turbines in icing conditions. Furthermore, it will help to reduce the uncertainties on the gains achieved through alternative operational strategies. The proposed methodology can be used by operators to compute the payback and quantify the added value of such modifications on their wind farm. It also helps to understand when a strategy is efficient, therefore helping to improve the development of new alternative operational strategies. Finally, using this methodology, through a case study, we demonstrated that the three control turbines could have recovered between 54 and 56% of their icing losses if they were equipped with a Borealis Wind Ice Protection System (IPS).

RipetaScore: Measuring the Quality, Transparency, and Trustworthiness of a Scientific Work.

A wide array of existing metrics quantifies a scientific paper's prominence or the author's prestige. Many who use these metrics make assumptions that higher citation counts or more public attention must indicate more reliable, better quality science. While current metrics offer valuable insight into scientific publications, they are an inadequate proxy for measuring the quality, transparency, and trustworthiness of published research. Three essential elements to establishing trust in a work include: trust in the paper, trust in the author, and trust in the data. To address these elements in a systematic and automated way, we propose the ripetaScore as a direct measurement of a paper's research practices, professionalism, and reproducibility. Using a sample of our current corpus of academic papers, we demonstrate the ripetaScore's efficacy in determining the quality, transparency, and trustworthiness of an academic work. In this paper, we aim to provide a metric to evaluate scientific reporting quality in terms of transparency and trustworthiness of the research, professionalism, and reproducibility.

Macrozoobenthic community assemblage as key indicator for mangrove restoration success in North Sumatra and Aceh, Indonesia

The recognition of the high value of mangrove forests and the wide array of ecosystem services they provide has motivated investment in worldwide restoration efforts. However, current metrics of functional restoration (other than seedling survival rates and plant community composition) are often not readily available for local community managers, highlighting an urgency to identify easy‐to‐measure indicators to assess the functionality of restored mangroves. The macrozoobenthic community could be such practical indicator, as macrozoobenthic communities are sensitive to changes in their environment, and can be easily surveyed within local managing programs. Focusing on three main mangrove management conditions (natural, planted, and naturally regenerated) in North Sumatra and the province of Aceh, Indonesia, we compared vegetation and macrozoobenthic community diversity indices and identified environmental variables that best describe the forest management conditions and their associated macrozoobenthic community assemblage. Results showed that community assemblage, rather than macrozoobenthic diversity index, was associated with management conditions. The highest dissimilarity in macrozoobenthic community assemblages occurred between planted versus natural mangroves, with nonsignificant dissimilarity between natural and naturally regenerated mangroves. The Lined Nerite gastropod (Nerita balteata) was identified as an indicator of natural mangroves, and the invasive Giant African snail (Achatina fulica) was abundant in mangrove plantations, but also in natural mangroves bordering harbors, oil palm plantations, and aquaculture ponds, suggesting associated anthropogenic pressures. This study showed that the macrozoobenthic community can be used as restoration indicator and could serve as a baseline to empower monitoring activities and community‐based adaptive management practices to improve the outcomes of restoration efforts.

Valuation: A New Approach to Measure the Performance of Last Planner System

The Last Planner System (LPS) is a production planning and control system that utilizes the lean construction philosophy to improve workflow reliability. Although there are several metrics developed to measure the performance of LPS, the current metrics fail to genuinely reflect the actual performance of LPS. The literature review shows no evidence of existing researches that provide a holistic approach to measure the performance of LPS. Consequently, this research aims to propose a new holistic approach to measure the performance of LPS called valuation. The objectives of this research are: (1) to provide an overview of the current metrics used to measure the performance of LPS, (2) to propose 10 steps to measure the performance of LPS based on the valuation approach, (3) to utilize the valuation approach to derive numerous metrics based on all possible relationships between the main and sub (i.e., activity and constraint) categories of LPS that currently available or may emerge in the future. This research contributes to the body of knowledge by deriving metrics based on the valuation approach that are significantly more comprehensive and mathematically more robust since they integrate several criteria and rely on the value rather than the number or amount of activities or constraints. Hence, the valuation approach generates more accurate results. Moreover, the valuation approach can help the construction professionals to track the performance of LPS across phases or even projects by accumulating the data and measuring the proposed metrics.

Learning-Based Quality Assessment for Image Super-Resolution

Image Super-Resolution (SR) techniques improve visual quality by enhancing the spatial resolution of images. Quality evaluation metrics play a critical role in comparing and optimizing SR algorithms, but current metrics achieve only limited success, largely due to the lack of large-scale quality databases, which are essential for learning accurate and robust SR quality metrics. In this work, we first build a large-scale SR image database using a novel semi-automatic labeling approach, which allows us to label a large number of images with manageable human workload. The resulting SR Image quality database with Semi-Automatic Ratings (SISAR), so far the largest of SR-IQA database, contains 12 600 images of 100 natural scenes. We train an end-to-end Deep Image SR Quality (DISQ) model by employing two-stream Deep Neural Networks (DNNs) for feature extraction, followed by a feature fusion network for quality prediction. Experimental results demonstrate that the proposed method outperforms state-of-the-art metrics and achieves promising generalization performance in cross-database tests. The SISAR database and DISQ model will be made publicly available to facilitate reproducible research.

Interplay Between γ–Ray Irradiation and 3DEG for Dosimeter Applications

This work investigates the cumulative dose ⁶⁰Co gamma (<inline-formula> <tex-math notation="LaTeX">$\gamma$ </tex-math></inline-formula>) &#x2013; ray irradiation effects on enhancement mode HEMT devices inheriting 3D &#x2013; Electron and Hole Gases for dosimeter applications. The devices are irradiated through a ⁶⁰Co source and demonstrate the enhancement in the drain current metrics. To elucidate, the said devices were irradiated through different mechanisms and the Compton effect was investigated through contour plots via TCAD simulations. The degradation of Schottky Gate contact and insulator charging affects the 2D &#x2013; Hole Gas at the GaN cap and thereby affects the bottleneck at the 3DEG sheet. This significantly affects the OFF &#x2013; state leakage components of the said device and can therefore be exploited for potential use in sensing and dosimeter applications. The leakage components can be exploited further to improve the linearity of the dosimeter by considering different grading profiles of the AlGaN layer. In this regard, a workflow for optimizing the sensitivity and linearity of the dosimeter through different graded profiles is also presented. Amongst all, gaussian graded profiles have been identified as the best &#x2013; case scenario considering the sensitivity and exhibiting a linear operation.

Cognitive Complexity Beyond Generalization: A Subjective Rating for the Human Comprehension

The cognitive complexity of a software determines the comprehension effort of a particular individual faces when designing, developing and maintaining a software. The comprehension level tends to be varied with each human resulting the cognitive complexity a subjective measurement. Expressing the cognitive complexity as a form of metric quantifies the comprehensibility as a generic value, which does not imply the subjectivity of human factor. This study elaborates the significance of expressing the cognitive complexity as a form of a subjective rating. The cognitive complexity rating has been pioneered with respect to the human and programming dependent factors related to human cognition. The Divisive hierarchical clustering algorithm has been used to train and predict the cognition rating per user. It has been clearly elaborated the subjectivity of the cognitive ratings over the quantitative and static complexity values of current cognitive and software complexity metrics. Thereby, the concept of cognition rates has been proposed as a preliminary step of determining and expressing the cognitive complexity.

Batch Evaluation Metrics in Information Retrieval: Measures, Scales, and Meaning

A sequence of recent papers, including in this journal, has considered the role of measurement scales in information retrieval (IR) experimentation, and presented the argument that (only) uniform-step interval scales should be used. Hence, it has been argued, well-known metrics such as reciprocal rank, expected reciprocal rank, normalized discounted cumulative gain, and average precision, should be either discarded as measurement tools, or adapted so that their metric values lie at uniformly-spaced points on the number line. These papers paint a rather bleak picture of past decades of IR evaluation, at odds with the IR community’s overall emphasis on practical experimentation and measurable improvement. Our purpose in this work is to challenge that pessimistic assessment. In particular, we argue that mappings from categorical and ordinal data to sets of points on the number line are valid provided there is an external reason for each target point to have been selected. We first consider the general role of measurement scales, and of categorical, ordinal, interval, ratio, and absolute data collections. In connection with the first two of those categories we also provide examples of the knowledge that is captured and represented by numeric mappings to the real number line. Focusing then on information retrieval, we argue that document rankings are categorical data, and that the role of an effectiveness metric is to provide a single value that summarizes the usefulness to a user or population of users of any given ranking, with usefulness able to be represented as a continuous variable on a ratio scale. That is, we argue that most current IR metrics are well-founded, and, moreover, that those metrics are more meaningful in their current form than in the proposed “intervalized” versions.

A Novel Algorithmic Similarity Measure for Collaborative Filtering: A Recommendation System Based on Rating Distances

Internet sources contain a vast amount of information about items that people desire to purchase. It is impossible to evaluate these resources and come to an informed decision. People need automated systems that evaluate previous information and propose item alternatives. Recommending items using a smart system, which is based on the previous user preferences, has growing importance since the available product data is exponentially growing. Additionally, it is difficult to find new and correct things that a user would like among this massive amount of data. To make accurate recommendations with a smart system, researchers and practitioners use collaborative filtering methods with similarity calculation based on user preferences. The crucial point in collaborative filtering is to find a valuable measure that resembles correct similarity between users. The current similarity metrics in the literature have some disadvantages in conducting accurate recommendations. To improve the recommendation performance, this study proposes a novel similarity measure that assesses the distance between the user’s ratings and the median score. Considering distance from the median score is essential since some users may prefer to rate close to the median rather than the extremes. Experiments were conducted with a famous collaborative filtering dataset. Results showed that proposed similarity measure demonstrated superior performance regarding the recommendation accuracy. Implications of our results for XYZ are discussed.

Association of readmission penalty amount with subsequent 30-day risk standardized readmission and mortality rates among patients hospitalized with heart failure: An analysis of get with the guidelines – heart failure participating centers

The Hospital Readmissions Reduction Program penalizes hospitals with excess 30-day risk-standardized readmission rates (RSRR) for heart failure (HF). The association of financial penalty amount with subsequent short-term clinical outcomes is unknown. Patients admitted to American Heart Association Get With The Guidelines-HF registry participating centers from October 1, 2012 through December 1, 2015 who had Medicare-linked data were included. October 2012 hospital-specific penalty amounts were calculated based on diagnosis-related group payments and excess readmission ratios. Adjusted Cox models were created to evaluate the association of penalty amount categories (non-penalized: 0%; low-penalized: >0%-<0.50%; mid-penalized ≥0.50%-<0.99%; high-penalized ≥0.99%) with subsequent 30-day RSRR and risk-standardized mortality rates (RSMR). Trends in post-discharge 30-day RSRR and RSMR from 2012 to 2015 were analyzed across hospitals stratified by penalty amount categories. The present study included 61,329 patients who were admitted across 262 hospitals. Compared with patients admitted to non-penalized hospitals (36.3%), those admitted to increasingly penalized hospitals were more likely to have higher 30-day RSRR (low-penalized [43.9%]: HR, 1.10 [95% CI, 1.04-1.16]; mid-penalized [12.0%]: HR, 1.07 [95% CI, 0.99-1.16]; high-penalized [7.9%]: HR, 1.23 [95% CI, 1.12-1.35]) but not 30-day RSMR. Over time, 30-day RSRR and RSMR did not meaningfully change across penalized versus non-penalized hospitals. Financial penalties based on 30-day RSRR are not associated with declines in 30-day RSRR or RSMR from 2012 to 2015 among patients hospitalized with HF. Financially penalizing hospitals based on current Hospital Readmissions Reduction Program metrics may not incentivize improvements in short-term clinical outcomes for HF.

Weisfeiler-Leman in the Bamboo: Novel AMR Graph Metrics and a Benchmark for AMR Graph Similarity

Abstract Several metrics have been proposed for assessing the similarity of (abstract) meaning representations (AMRs), but little is known about how they relate to human similarity ratings. Moreover, the current metrics have complementary strengths and weaknesses: Some emphasize speed, while others make the alignment of graph structures explicit, at the price of a costly alignment step. In this work we propose new Weisfeiler-Leman AMR similarity metrics that unify the strengths of previous metrics, while mitigating their weaknesses. Specifically, our new metrics are able to match contextualized substructures and induce n:m alignments between their nodes. Furthermore, we introduce a Benchmark for AMR Metrics based on Overt Objectives (Bamboo), the first benchmark to support empirical assessment of graph-based MR similarity metrics. Bamboo maximizes the interpretability of results by defining multiple overt objectives that range from sentence similarity objectives to stress tests that probe a metric’s robustness against meaning-altering and meaning- preserving graph transformations. We show the benefits of Bamboo by profiling previous metrics and our own metrics. Results indicate that our novel metrics may serve as a strong baseline for future work.

Kidney transplant program specific reporting and transplant metrics.

Kidney transplantation is a heavily regulated medical procedure with the Secretary of HHS ultimately responsible for oversight and authority derived from the NOTA and the Final Rule. Transplant Programs undergo publicly reported evaluations every 6 months based on outcomes from a 2-and-a-half-year period. The current Bayesian metrics for kidney transplant programs were created such that over ten percentage of programs are deemed underperformers, or 'flag', every 6 months. Newly suggested transplant metrics have been released for public comment in Summer 2021. In addition to graft outcomes, waiting list mortality and organ acceptance rate ratios are proposed. Under the newly proposed kidney transplant metrics, over 10% of programs are expected to be deemed underperformers or 'flagged'. Transplant Center flagging is well correlated with decreased transplantation due to the transplant centres move to more conservative organ and patient acceptance. Death on the waiting list is a proposed metric over which transplant centres have little influence. In the USA, the harsh regulation continued by Health Resources and Services Administration (HRSA) through the national organ procurement and transplant network (OPTN) and Scientific Registry for Transplant Recipients (SRTR) leads directly to high organ discard rates and limitations to transplanting patients with perceived unadjusted risks. Instead of loosening regulation in a highly functioning industry that achieves remarkable outcomes in end stage kidney patients, the OPTN with the SRTR persist in increasing potential penalties through more proposed metrics that continue to deem 10% of US kidney transplant programs as underperformers. HRSA must establish a reasonable regulatory environment that allows for innovation and increased transplant opportunities for US end-stage renal disease patients.

African Americans’ Hemodialysis Treatment Adherence Data Assessment and Presentation: A Precision-Based Paradigm Shift to Support Quality Improvement Activities

Rationale & ObjectiveThrice-weekly hemodialysis can result in adequate urea clearance; however, the morbidity and mortality rates of patients treated with maintenance dialysis remain unacceptably high, partly because of nonadherence. African Americans have a higher prevalence of kidney failure treated with dialysis, greater dialysis nonadherence, and higher odds of hospitalization. We hypothesized that more precise ways of assessing dialysis treatment adherence will reflect the severity of nonadherence, distinguish patterns of nonadherence, and inform the design of personalized behavioral interventions.Study DesignRetrospective cohort study.Setting & ParticipantsAfrican American patients receiving hemodialysis for >90 days.ExposureHemodialysis.OutcomeDialysis adherence.Analytical ApproachDialysis attendance data were displayed using a dot plot, categorized based on missed and shortened treatments, and examined for patterns. Descriptive characteristics were reported. In an exploratory analysis, associations between dialysis treatment adherence and participant characteristics were evaluated using ordinary least squares regression. An analysis was performed using missed minutes of dialysis and current metrics for measuring dialysis treatment adherence (ie, missed and shortened treatments).ResultsAmong 113 African American patients treated with dialysis, 47% were men; the median age was 57 years (interquartile range, 46-70 years), and the median dialysis vintage was 54 months (interquartile range, 22-90 months). With rows ordered based on the total missed minutes of dialysis, the dot plot displayed a decreasing gradient in the severity of nonadherence, with novel dialysis treatment adherence categories termed as follows: consistent underdialysis, inconsistent dialysis, and consistent dialysis. Distinct patterns of nonadherence and heterogeneity emerged within these categories. Older age was consistently associated with better adherence, as determined by the analyses performed using the total missed minutes of dialysis as well as missed and shortened treatments.LimitationsThe study findings, although replicable and paradigm-shifting, might be limited by the short timeline, focus on adherence data specific to African American patients treated with dialysis, and restriction to dialysis units affiliated with 1 academic center.ConclusionsThis study presents more precise and novel ways of measuring and displaying dialysis treatment adherence. The findings introduce a more personalized approach for evaluating actual dialysis uptake. Identification of unique patterns of adherence behavior is important to inform the design of effective behavioral interventions and improve outcomes for vulnerable African American patients treated with dialysis.

Investigation of Multi-Mesa-Channel-Structured AlGaN/GaN MOSHEMTs with SiO2 Gate Oxide Layer

In this study, an electron-beam lithography system was employed to pattern 80-nm-wide and 980-nm-spaced multi-mesa-channel for fabricating AlGaN/GaN metal-oxide-semiconductor high electron mobility transistors (MOSHEMTs). Since the structure of multi-mesa-channel could enhance gate control capabilities and reduce the self-heating effect in the channel, the performance of the MOSHEMTs could be obviously improved. The direct current performance metrics of the multi-mesa-channel-structured MOSHEMTs, such as a saturation drain-source current of 929 mA/mm, maximum extrinsic transconductance of 223 mS/mm, and on-resistance of 2.1 Ω-mm, were much better than those of the planar-structured MOSHEMTs. Moreover, the threshold voltage of the multi-mesa-channel-structured MOSHEMTs shifted toward positive voltage from −2.6 to −0.6 V, which was attributed to the better gate control capability. Moreover, the multi-mesa-channel-structured MOSHEMTs also had superior high-frequency and low-frequency noise performance. A low Hooge’s coefficient of 1.17 × 10−6 was obtained.

Quantifying dissimilarities between heterogeneous networks with community structure

Quantifying dissimilarities between networks is a fundamental and challenging problem in network science. Current metrics for network comparison either assume the homogeneous distribution of nodal degrees or ignore the community structure of the network. Here we propose an efficient measure for comparing heterogeneous networks with communities from the perspective of probability distribution functions, which incorporates the nodal distance distribution, the clustering coefficient distribution and the alpha centrality distribution. Comparison between community benchmarks shows that the proposed measure returns non-zero values only when the networks are non-isomorphic.

The generative capacity of probabilistic protein sequence models

Potts models and variational autoencoders (VAEs) have recently gained popularity as generative protein sequence models (GPSMs) to explore fitness landscapes and predict mutation effects. Despite encouraging results, current model evaluation metrics leave unclear whether GPSMs faithfully reproduce the complex multi-residue mutational patterns observed in natural sequences due to epistasis. Here, we develop a set of sequence statistics to assess the “generative capacity” of three current GPSMs: the pairwise Potts Hamiltonian, the VAE, and the site-independent model. We show that the Potts model’s generative capacity is largest, as the higher-order mutational statistics generated by the model agree with those observed for natural sequences, while the VAE’s lies between the Potts and site-independent models. Importantly, our work provides a new framework for evaluating and interpreting GPSM accuracy which emphasizes the role of higher-order covariation and epistasis, with broader implications for probabilistic sequence models in general.

How Good is Too Good? A Subjective Study on Over Enhancement of Images

For a long time different studies have focused on introducing new image enhancement techniques. While these techniques show a good performance and are able to increase the quality of images, little attention has been paid to how and when overenhancement occurs in the image. This could possibly be linked to the fact that current image quality metrics are not able to accurately evaluate the quality of enhanced images. In this study we introduce the Subjective Enhanced Image Dataset (SEID) in which 15 observers are asked to enhance the quality of 30 reference images which are shown to them once at a low and another time at a high contrast. Observers were instructed to enhance the quality of the images to the point that any more enhancement will result in a drop in the image quality. Results show that there is an agreement between observers on when over-enhancement occurs and this point is closely similar no matter if the high contrast or the low contrast image is enhanced.

Colourlab Image Database: Geometric Distortions

Over the years, a high number of different objective image quality metrics have been proposed. While some image quality metrics show a high correlation with subjective scores provided in different datasets, there still exists room for improvement. Different studies have pointed to evaluating the quality of images affected by geometrical distortions as a challenge for current image quality metrics. In this work, we introduce the Colourlab Image Database: Geometric Distortions (CID:GD) with 49 different reference images made specifically to evaluate image quality metrics. CID:GD is one of the first datasets which include three different types of geometrical distortions; seam carving, lens distortion, and image rotation. 35 state-ofthe-art image quality metrics are tested on this dataset, showing that apart from a handful of these objective metrics, most are not able to show a high performance. The dataset is available at &lt;ext-link ext-link-type="url" xlink:href="http://www.colourlab.no/cid"&gt;www.colourlab.no/cid&lt;/ext-link&gt;.

Deep Learning and Harmonization of Multi-Institutional Data for Automated Gross Tumor and Nodal Segmentation for Oropharyngeal Cancer

Automated tumor segmentation for oropharyngeal cancer (OPC) has the potential to improve treatment planning, response assessment, and clinical translation of imaging-based biomarkers. Deep learning has shown promise for cancer imaging segmentation, but performance for OPC tumors has been suboptimal with studies generally limited to small single-institution settings. In this study, we curated and harmonized multiple heterogeneous, multi-institutional datasets to develop and validate computed tomography (CT)-based, deep learning models for total gross tumor volume (GTV), primary (GTVp), and nodal (GTVn) segmentations.Data was obtained from The Cancer Imaging Archive (TCIA) and included 1228 CT simulation scans from OPC patients treated with definitive radiotherapy collected from 2003-2014 from four institutions that included original gross tumor volumes (GTV) as delineated by the treating radiation oncologist. GTVs were manually reviewed and ground-truth labels were harmonized to include distinct GTVp and GTVn labels. Cases were split randomly, such that 70% of cases were used for model training, 15% for tuning, and 15% for independent performance testing. Utilizing a modified 3D UNET-based architecture, models were trained and tuned to predict total GTV, GTVp and GTVn. Model performance was assessed by measuring precision, recall, and dice score coefficient (DSC) when comparing deep learning-generated to ground truth volumes in the independent test set. Patients without contrast-enhanced CT were excluded from the GTVp model a priori, given the importance of contrast in delineating primary tumor.Algorithms' median performance on test sets with 95% confidence intervals. GTVp test set limited to contrast-enhanced scans. Within the total dataset (n = 1228), median age was 59, cases were 82% male, and HPV status was 49% positive, 16% negative, and 35% unknown. Tumor staging was 45% T3-4, 54% T1-T2, and 1% T0/X. Nodal staging was 75% N2-N3 and 25% N0-1. Overall, the GTVn model had the highest performance (median DSC: 0.76; 95% CI: 0.73, 0.77), followed by total GTV (median DSC: 0.71; 95% CI: 0.68, 0.73), and GTVp (median DSC 0.68; 95% CI: 0.63, 0.71).Deep learning with multiple harmonized data sources can yield effective models for OPC primary and nodal segmentation using CT alone. The utility of these models will depend on the clinical use case and will be explored on further investigation, though current model performance metrics, particularly for nodal segmentation, are likely adequate for prospective testing in clinical and research applications.