Parametric Distribution Research Articles

Composite and Mixture Distributions for Heavy-Tailed Data—An Application to Insurance Claims

This research provides a comprehensive analysis of two-component non-Gaussian composite models and mixture models for insurance claims data. These models have gained attraction in actuarial literature because they provide flexible methods for curve-fitting. We consider 256 composite models and 256 mixture models derived from 16 popular parametric distributions. The composite models are developed by piecing together two distributions at a threshold value, while the mixture models are developed as convex combinations of two distributions on the same domain. Two real insurance datasets from different industries are considered. Model selection criteria and risk metrics of the top 20 models in each category (composite/mixture) are provided by using the ‘single-best model’ approach. Finally, for each of the datasets, composite models seem to provide better risk estimates.

Taking cues from machine learning, compartmental and time series models for SARS-CoV-2 omicron infection in Indian provinces

SARS-CoV-2, the virus responsible for COVID-19, posed a significant threat to the world. We analyzed COVID-19 dissemination data in the top ten Indian provinces by infection incidences using the Susceptible-Infectious-Removed (SIR) model, an Autoregressive Integrated Moving Average (ARIMA) time series model, a machine learning model based on the Random Forest, and distribution fitting. Outbreaks are expected to continue if the Basic Reproduction Number (R0) > 1, and infection waves are anticipated to end if the R0 < 1, as determined by the SIR model. Different parametric probability distributions are also fitted. Data collected from December 12, 2021, to March 31, 2022, encompassing data from both before and during the implementation of strict control measures. Based on the estimates of the model parameters, health agencies and government policymakers can develop strategies to combat the spread of the disease in the future, and the most effective technique can be recommended for real-world application for other outbreaks of COVID-19. The best method out of these could be also implemented further on the epidemiological data of other similar infectious agents.

Accounting for hydroclimatic properties in flood frequency analysis procedures

Abstract. Flood hazard is typically evaluated by computing extreme flood probabilities from a flood frequency distribution following nationally defined procedures in which observed peak flow series are fit to a parametric probability distribution. These procedures, also known as flood frequency analysis, typically recommend only one probability distribution family for all watersheds within a country or region. However, large uncertainties associated with extreme flood probability estimates (&gt;50-year flood or Q50) can be further biased when fit to an inappropriate distribution model because of differences in the tails between distribution families. Here, we demonstrate that hydroclimatic parameters can aid in the selection of a parametric flood frequency distribution. We use L-moment diagrams to visually show the fit of gaged annual maxima series across the United States, grouped by their Köppen climate classification and the precipitation intensities of the basin, to a general extreme value (GEV), log normal 3 (LN3), and Pearson 3 (P3) distribution. Our results show that in real space basic hydroclimatic properties of a basin exert a significant influence on the statistical distribution of the annual maxima. The best-fitted family distribution shifts from a GEV towards an LN3 distribution across a gradient from colder and wetter climates (Köppen group D, continental climates) towards more arid climates (Köppen group B, dry climates). Due to the diversity of hydrologic processes and flood-generating mechanisms among watersheds within large countries like the United States, we recommend that the selection of distribution model be guided by the hydroclimatic properties of the basin rather than relying on a single national distribution model.

Perspective study on charge time measurement of long-term stored lithium-ion batteries used in electric-powered aircraft assessed and modelled by specific growth model with diffusion process backup

Electrical energy sources and batteries are currently in great demand. The demand is growing, and their use is on the rise in a variety of applications. In contrast, there are limited both production/manufacturing options and an effort to ensure more environmentally friendly acquisition, operation and use. In this article, we study, test, and model the charging process of Li-ION batteries. We study a set of long-term stored Li-ION batteries and compare the data and results with a set of new Li-ION batteries. The charging data are obtained experimentally by long-term testing, and the volume (of the data) is very large and statistically very representative. Modelling of the Li-ION battery charging process is based on specific growth model forms. Furthermore, the charging process is simulated using special forms of the diffusion process, specifically the Geometric Ornstein-Uhlenbeck process, which seems to be the most suitable for this purpose. The research is carried out with the aim to obtain the first passage time (FPT), i.e. the charge time of the battery. The first passage time is further studied with respect to the suitability of its expression using parametric probability distributions. We also acquired the numerical statistics of the charging time FPT as well as the respective parametric distribution. The FPT results obtained for long-term stored and new Li-ION batteries were compared using well-known and lesser-known statistical tests, such as the Kullback-Leibler divergence.

Bayesian multinomial model with inference based on Pólya-Gamma augmentation

Statistical inferences from multi-categorical data are often encountered in many scientific fields. It is customary in the literature to model the probability of occurrence of a certain category on some risk factors. However, this work presents a framework to model the stick-breaking probabilities of a multi-categorical event on risk factors. In addition, we adopted a Pólya-Gamma data augmentation to arrive at a conjugate prior distribution for the model parameters and developed a cheap Gibbs sampling algorithm for model estimation. The proposed model allows for flexible and computationally efficient modeling of the conditional properties of an ordered categorical process and the incorporation of dependencies between the different categories of the outcome. The framework was used to model data derived from the Nigeria Demographic and Health Survey to study the spatial patterns and risk factors of maternal malnutrition in Nigeria. The findings identified the determinant factors and showed heterogeneity in the spatial patterns of maternal malnutrition prevalence across Nigeria. Findings from the analysis could be relevant for designing intervention programs and guiding policy-making, such as in the allocation of limited resources across the country.

Sensitivity analysis of prior distributions in Bayesian graphical modeling: Guiding informed prior choices for conditional independence testing

Bayesian analysis methods provide a significant advancement in network psychometrics, allowing researchers to use the edge inclusion Bayes factor for testing conditional independence between pairs of variables in the network. Using this methodology requires setting prior distributions on the network parameters and on the network’s structure. However, the impact of both prior distributions on the inclusion Bayes factor is underexplored. In this paper, we focus on a specific class of Markov Random Field models for ordinal and binary data. We first discuss the different choices of prior distributions for the network parameters and the network structure, and then perform an extensive simulation study to assess the sensitivity of the inclusion Bayes factor to these distributions. We pay particular attention to the effect of the scale of the prior on the inclusion Bayes factor. To improve the accessibility of the results, we also provide an interactive Shiny app. Finally, we present the R package simBgms, which provides researchers with a user-friendly tool to perform their own simulation studies for Bayesian Markov Random Field models. All of this should help researchers make more informed, evidence-based decisions when preparing to analyze empirical data using network psychometric models.

A Review on Parametric and Semiparametric Distributions in Characterizing Synthetic Aperture Radar Clutter Data

A Review on Parametric and Semiparametric Distributions in Characterizing Synthetic Aperture Radar Clutter Data

Use of random integration to test equality of high dimensional covariance matrices.

Testing the equality of two covariance matrices is a fundamental problem in statistics, and especially challenging when the data are high-dimensional. Through a novel use of random integration, we can test the equality of high-dimensional covariance matrices without assuming parametric distributions for the two underlying populations, even if the dimension is much larger than the sample size. The asymptotic properties of our test for arbitrary number of covariates and sample size are studied in depth under a general multivariate model. The finite-sample performance of our test is evaluated through numerical studies. The empirical results demonstrate that our test is highly competitive with existing tests in a wide range of settings. In particular, our proposed test is distinctly powerful under different settings when there exist a few large or many small diagonal disturbances between the two covariance matrices.

FocalPose++: Focal Length and Object Pose Estimation Via Render and Compare.

We introduce FocalPose++, a neural render-and-compare method for jointly estimating the camera-object 6D pose and camera focal length given a single RGB input image depicting a known object. The contributions of this work are threefold. First, we derive a focal length update rule that extends an existing state-of-the-art render-and-compare 6D pose estimator to address the joint estimation task. Second, we investigate several different loss functions for jointly estimating the object pose and focal length. We find that a combination of direct focal length regression with a reprojection loss disentangling the contribution of translation, rotation, and focal length leads to improved results. Third, we explore the effect of different synthetic training data on the performance of our method. Specifically, we investigate different distributions used for sampling object's 6D pose and camera's focal length when rendering the synthetic images, and show that parametric distribution fitted on real training data works the best. We show results on three challenging benchmark datasets that depict known 3D models in uncontrolled settings. We demonstrate that our focal length and 6D pose estimates have lower error than the existing state-of-the-art methods.

Valorization of bio-oil distillation residue through co-pyrolysis with moso bamboo to prepare biochar: Optimization study on effects of blend ratio, pyrolysis temperature, and residence time

In view of the high energy consumption and pollutant emissions in the cement production industry, biomass as the renewable carbon-neutrality energy is promisingly conducive to remitting cement carbon dioxide emissions and simultaneously realizing waste-to-energy initiatives. In this study, bio-oil distillation residue (DR), with high carbon content and heat value, is co-pyrolyzed with moso bamboo (MB) to optimize the bulk and combustion properties of biochar, aiming at yielding bio-fuel as an alternative fuel to replace coals for energy and heat supply in cement industry. The impacts of three primary process parameters on the carbonization reaction were investigated, while the parametric interactions were further estimated by using response surface methodology (RSM) to optimize the ultimate, proximate and fuel characteristics of biochars. The results showed the co-pyrolysis synergies between bio-oil distillation residue and moso bamboo endowed the maximum biochar yield of 44.62% and the maximum difference biochar yield of 7.56%. Meanwhile, the experimental carbon content, fixed carbon content and high heat value (HHV) were regulated through coupling parametric distribution and further increased by 1.96%, 2.59% and 1.23 MJ/kg, respectively. The parametric responses on combustion characteristics of biochar based on thermogravimetric analysis were predicted, and the optimal carbonization conditions for biochar production were also summarized. This study would offer an optimal biochar production scheme to adaptively yield coal-like biochar for reducing anthropogenic carbon dioxide emissions in cement industry.

Bayesian random-effects meta-analysis with empirical heterogeneity priors for application in health technology assessment with very few studies.

In Bayesian random-effects meta-analysis, the use of weakly informative prior distributions is of particular benefit in cases where only a few studies are included, a situation often encountered in health technology assessment (HTA). Suggestions for empirical prior distributions are available in the literature but it is unknown whether these are adequate in the context of HTA. Therefore, a database of all relevant meta-analyses conducted by the Institute for Quality and Efficiency in Health Care (IQWiG, Germany) was constructed to derive empirical prior distributions for the heterogeneity parameter suitable for HTA. Previously, an extension to the normal-normal hierarchical model had been suggested for this purpose. For different effect measures, this extended model was applied on the database to conservatively derive a prior distribution for the heterogeneity parameter. Comparison of a Bayesian approach using the derived priors with IQWiG's current standard approach for evidence synthesis shows favorable properties. Therefore, these prior distributions are recommended for future meta-analyses in HTA settings and could be embedded into the IQWiG evidence synthesis approach in the case of very few studies.

A contagion test with unspecified heteroscedastic errors

The tests of contagion in Fry-McKibbin et al. (2010) filter returns by a vector autoregressive model, assume residuals are independent, fit a parametric distribution family to residuals, and test for the change of contagion measures, which ignore the effect of filtering the time series model and the stylized fact of heteroscedasticity in daily returns. This paper studies a contagion test based on correlation by allowing heteroscedastic errors with a deterministic jump from the pre-crisis to the crisis periods. Because the developed test does not infer heteroscedasticity, it is robust against heteroscedasticity but its asymptotic variance under the null hypothesis of no contagion becomes complicated, which relies on a block method for estimating the asymptotic variance. A simulation study confirms the good finite sample performance of the new contagion test. Finally, we apply the test to three datasets to test for contagion.

OPTIMIZATION OF THE AUCTION DURATION IN THE PRESENSE OF TIME-DEPENDANT COSTS

This paper examines the influence of the duration of auctions or tenders on the expected gain of their organizer. Extending the duration of bidding affects auction results in two ways. On the one hand, it allows attracting a larger number of participants to the auction, and the competition between them increases the chances of the auctioneer to get a better price. On the other hand, delaying bids delays the receipt of money (for auctions) or required goods or services (for tenders), and time has value in itself. The influence of these two factors, which act in opposite directions, suggests the existence of an optimal duration of the bidding process. The paper develops a mathematical model of bidding, which formalizes these considerations and provides an algorithm to determine their optimal duration. The arrival of bidders willing to participate in the auction is modeled as a Poisson process. Each participant is characterized by his own assessment of the value of the object put up for auction. These estimates are assumed to be independent identically distributed random variables drawn from some parametric distribution. Under these assumptions, Myerson's revenue equivalence theorem makes it possible to predict the expected results of the auction as a function of the number of bidders, regardless of the auction rules. On this basis, it is possible to compare the benefits and costs associated with changing the duration of time for accepting applications for participation in bidding, which makes it possible to determine its optimal value. The obtained optimality conditions have a meaningful and intuitive economic interpretation. For practical applications, the use of Monte Carlo methods based on the empirical distribution of bid and ask prices is proposed. The practical implementation of the proposed algorithm can improve the economic performance of the auctioneer, which is especially relevant for the public sector of the economy.

Parametric Information Geometry with the Package Geomstats

We introduce the information geometry module of the Python package Geomstats. The module first implements Fisher–Rao Riemannian manifolds of widely used parametric families of probability distributions, such as normal, gamma, beta, Dirichlet distributions, and more. The module further gives the Fisher–Rao Riemannian geometry of any parametric family of distributions of interest, given a parameterized probability density function as input. The implemented Riemannian geometry tools allow users to compare, average, interpolate between distributions inside a given family. Importantly, such capabilities open the door to statistics and machine learning on probability distributions. We present the object-oriented implementation of the module along with illustrative examples and show how it can be used to perform learning on manifolds of parametric probability distributions.

Improved estimation of overall survival and progression-free survival for state transition modeling.

Aim: National Institute for Health and Care Excellence guidance (Technical Support Document 19) highlights a key challenge of state transition models (STMs) being their difficulty in achieving a satisfactory fit to the observed within-trial endpoints. Fitting poorly to data over the trial period can then have implications for long-term extrapolations. A novel estimation approach is defined in which the predicted overall survival (OS) and progression-free survival (PFS) extrapolations from anSTM are optimized to provide closer estimates of the within-trial endpoints. Materials & methods: An STM was fitted to the SQUIRE trial data in non-small-cell lung cancer (obtained from Project Data Sphere). Two methods were used: a standard approach whereby the maximum likelihood was utilized for the individual transitions and the best-fitting parametric model selected based on AIC/BIC, and a novel approach in which parameters were optimized by minimizing the area between the STM-predicted OS and PFS curves and the corresponding OS and PFS Kaplan-Meier curves. Sensitivity analyses were conducted to assess uncertainty. Results: The novel approach resulted in closer estimations to the OS and PFS Kaplan-Meier for all combinations of parametric distributions analyzed compared with the standard approach. Though the uncertainty associated with the novel approach was slightly larger, it provided better estimates to the restricted mean survival time in 10 of the 12 parametric distributions analyzed. Conclusion: A novel approach is defined which provides an alternative STM estimation method enabling improved fits to modeled endpoints, which can easily be extended to more complex model structures.

Bayesian Risk Assessment Technique for Economic Stress-Strength Models

This paper explores two areas of risk assessment modelling in economics and business: the Stress-Strength model and Bayesian techniques. The model assumes that the probability of stress exceeding strength is a measure of risk. The interpretation of stress and strength largely depends on the particular event or phenomenon being modelled. The use of the Stress-Strength model is demonstrated through the Gaussian assumption of probability distributions for random model parameters, particularly in assessing the risk of not achieving a required margin value. The concept of the capability function, representing the difference between strength and stress, is introduced in the modelling process. The probability distribution for the capability function is initially determined based on the Gaussian distribution of the random variables used in the model, allowing for evaluating the risk metric. The Bayesian approach is then applied to generalise the problem statement when dealing with unknown parameters of probability distributions for the Stress and Strength models. The uncertainty of these parameters is modelled through uniform probability distributions, and equations for calculating prior and posterior estimates are consistently obtained. Since multidimensional integrals are involved in these calculations, and solutions cannot be obtained in closed analytical form, Monte Carlo simulation is used to solve this computation problem.

Robust Statistical Modeling of Heterogeneity for Repairable Systems Using Multivariate Gaussian Convolution Processes

A main challenge in reliability analysis of repairable systems is to model the heterogeneity in their failure behavior, which can be reflected by the corresponding recurrent failure-time data. To capture the system heterogeneity for data analysis, a system-specific random effect is typically introduced in most existing statistical models. In practice, the random effect of repairable systems tends to be time varying; for example, each repair action could change system's physical properties. Prior studies, however, generally do not take account of this time-varying nature and few of them circumvent the risk of model misspecification on the parametric distribution of frailty. In this article, we propose a semiparametric model that uses multivariate Gaussian convolution processes (MGCPs) to meet the above challenges. First, we use the trend RP to model the baseline intensity function of each repairable system. Based on the baseline intensity function, we then introduce MGCPs to simultaneously factor in heterogeneity and infer commonalities across multiple systems. A Bayesian framework is used for parameter estimation and time-to-failure prediction. Simulation studies show the advantages of our model in terms of robustness and estimation accuracy. A group of oil and gas well systems are used to illustrate the application of the proposed model.

Preliminary clinical study of mucoadhesive propolis gel to gingival disorders control in patients with fixed orthodontic appliances.

Aims: Gingival hyperplasia (GH) etiology is diverse and ranges from acute or chronic inflammatory processes and is commonly observed in patients with fixed orthodontic appliances. Propolis has been globally studied and has anti -inflammatory and antimicrobial properties, however, there are few clinical studies involving orthodontics. Study design: This clinical study aims to obtain preliminary evidence of an oral gel containing 5% Brazilian propolis (BPG) in GH and gingivitis control on patients with fixed orthodontic appliances. Place and Duration study: Clinic Orthodontics, Faculty of Dentistry/UFMG, Belo Horizonte, Brazil, from March to October 2013. Methodology: Thirty patients were selected, but only 10 participated and reached the end of the study. The patients were submitted to questionnaire and a complete prophylaxis of dental structures before initiating the gel use(T0) and they were instructed to use the gel two times daily after brushing (morning and evening) during 30 days (T30). Patients were evaluated considering plaque and gingival index and photographs collection of first premolars - measuring gingiva of the clinical crown area. The SPSS version 18.0 statistical program, Shapiro Wilk normality test, presenting a normal or parametric distribution and all analyzes were performed using Student's t Test for paired samples. Results: BPG show statistically significant reduction of GBI and GSI (GB1 p =0.207; GB2 p = 0.122), respectively. Conclusion: Considering this study model, 5% propolis gel had effective in reducing GH and gingivitis. However, a double blind and randomized clinical studies with higher number of patients should be made to give greater credibility for gel use.

Differential Host Response Contributing to Incidence of Arterial and Venous Thromboembolism at Index Hospitalization for Sars-Cov-2 Infection across Wuhan, Delta, Omicron and Omicron BA.x Variants

Introduction SARS-CoV-2 induced coagulopathy has been associated with increased risk of arterial and venous thromboembolism. Literature demonstrates that incidence of thromboembolism has varied widely throughout the COVID-19 pandemic, with many reports preceding emergence of Delta and Omicron variants. This study aimed to characterize the incidence of arterial and venous thromboembolic events among patients hospitalized with acute SARS-CoV-2 infection during Wuhan, Delta, Omicron, or Omicron BA.x waves. To our knowledge, this study debuts variant specific analyses to shed some light on reported substantive differences pooled across SARS-CoV-2 variants regarding host response, clinical course, and outcomes. Intensity of pathophysiologic anomalies may have been associated with macro- and micro-circulatory clotting. Therefore, we hypothesized variant idiosyncratic host responses would be evoked and aimed to link our patient's demographics and comorbidity patterns with risk of thromboembolism. Methods This is a retrospective study including patients with laboratory confirmed SARS-CoV-2 infection who underwent index hospitalization during Wuhan (March 14, 2020- June 18, 2021), Delta (June 19, 2021- December 18, 2021), Omicron (December 19-March 30, 2022) or Omicron BA.x (March 31, 2022 - April 14, 2023) variant waves at an 820-bed academic public health trust hospital in Florida. Demographics, laboratory at presentation, ICD-10-CM-based comorbidity, SARS-CoV-2 directed treatment and administrative data were extracted under IRB exemption from electronic medical records. Generalized regression with adaptive LASSO identified variables associated with embolism in at least one variant wave controlling for extant dysrhythmia, chronic anticoagulation therapy, and COVID-19 directed treatment. Prespecified 2-way ANOVA examined interactions between intra- vs. inter-variant waves but was ruled out due to strong imbalances in variation. Thus, only within-variant contrasts were performed. Continuous data summarized with mean [±SD] were compared using an ANOVA test adjusting p-value for parametric vs. non-parametric distribution. Discrete data summarized as proportions were compared with chi-squared test. A p&amp;lt;.05 was considered significant. Results We studied 6490 patients consecutively discharged across Wuhan (n=2249), Delta (n=1196), Omicron (n=953) and Omicron BA.x (n=2092) variants with respective embolic event(s) of 3%, 2%, 2%, and 4%. Table 1 summarizes demographics, anthropometric, comorbidities and laboratory biomarkers obtained within 24h of hospitalization across variants. Table 2 presents clinical course and outcomes. Discussion Incidence of embolic events was isomorphically distributed among waves tending to affect older males harboring hypertension, neurologic disease, and/or heart failure. D-dimer levels at presentation significantly distinguished risk of thromboembolism with Wuhan, Delta, or Omicron, but not Omicron BA.x variant infection. Indeed, Omicron BA.x infection was associated with numerically lower ICU visits, mechanical ventilation requirement, hospital LOS and mortality. Elevated D-dimer consistently has been associated with excess risk of progression to severe illness and mortality (Milenkovic M et al., 2022), mostly via pooled results across SARS-CoV variant waves. Some have suggested that D-dimer guided anticoagulation management improves outcomes in patients with COVID-19 (Ronderos Botero DM et al., 2023). Deeper phenotyping of patients infected with Omicron BA.x variants may reveal unique prothrombotic features driving what appears to be a less severe hypercoagulable state.

Glofitamab Results in Cost Savings Versus Epcoritamab in Diffuse Large B-Cell Lymphoma (DLBCL): A Total Cost of Care Analysis

Background: Bispecific CD20xCD3 T-cell engaging antibodies, including glofitamab and epcoritamab, are a new class of treatment for DLBCL, with recent Food and Drug Administration approvals in the United States (US). Given the differences between glofitamab versus epcoritamab, respectively, with regards to 1) mode of administration (intravenous vs subcutaneous), 2) treatment course (12-cycle fixed-duration treatment vs treat to progression), 3) frequency of administration per cycle, and 4) drug acquisition costs, we compared the total costs of care for the two drugs across several time horizons. Methods: Per-patient total cost of care was examined for glofitamab versus epcoritamab across various cost categories. Drug costs were based on the Wholesale Acquisition Cost reported in AnalySource ® 2023 and treatment administration costs were based on the Centers for Medicare &amp; Medicaid Services physician fee schedule from 2023. Costs arising from adverse reactions, including all-grade cytokine release syndrome, were obtained from the Healthcare Cost and Utilization Project Database (2017) or from Badarocco et al. Transplant Cell Ther 2023. Routine care costs were acquired from Tkacz et al. Leuk Lymphoma 2020. The analysis was conducted from a US healthcare perspective. Subsequent treatment costs following the glofitamab or epcoritamab treatment regimens were not included. Glofitamab treatment costs were based on time-to-off-treatment Kaplan-Meier curves from the pivotal Phase II trial (NP30179; NCT03075696), which take into account discontinuation due to toxicity and progression, and epcoritamab treatment costs were based on progression free survival curves from Thieblemont et al. J Clin Oncol 2023 (EPCORE NHL-1; NCT03625037), extrapolated using parametric distribution methods. Glofitamab is administered intravenously every 21 days until disease progression or to a maximum of 12 cycles. Epcoritamab is administered subcutaneously every 28 days until disease progression. Per-patient total and incremental cumulative cost savings were estimated over 24 treatment cycles, and over 1, 5 and 10 years, as well as lifetime. We examined cost differences during Cycles 1-3 (when dose administrations for epcoritamab are the most frequent at 4 injections per cycle). As available overall survival data from both pivotal trials are still likely immature, the model used a common progression/survival probability for both treatments to minimize uncertainty in the estimation of total costs over time. Costs were adjusted to 2023 US dollars using the Consumer Price Index and included 3% discounting. Results: Total costs per patient were lower for glofitamab versus epcoritamab, resulting in cost savings at every cumulative cycle ( Figure A). The cost saving per patient with glofitamab versus epcoritamab across Cycles 1-3, when epcoritamab has the most injections per cycle, was $56,275 ($95,904 for glofitamab vs $152,179 for epcoritamab). Per-patient cost savings were also observed for glofitamab versus epcoritamab over all time horizons, including 1 year ($68,195), 5 years ($223,692), 10 years ($325,175), and over the lifetime ($503,075) ( Figure B). Per-patient costs arising from drug treatments were consistently lower for glofitamab compared with epcoritamab ( Figure B). Per-patient incremental adverse reaction costs ($364) and per-patient incremental treatment administration costs ($8,398) were higher for glofitamab versus epcoritamab, respectively, over 1 year. Conclusions: Glofitamab results in per-patient cost savings compared with epcoritamab at every cumulative administration cycle, particularly in the first few cycles when epcoritamab has more frequent dosing per cycle than glofitamab. Furthermore, glofitamab has lower total costs across all time horizons examined in this study (1-year, 5-year, 10-year, and lifetime). The lower total costs with glofitamab can be attributed to 1) lower annual drug acquisition costs, 2) fixed-duration treatment with a maximum of 12 cycles, and 3) less frequent dosing in earlier cycles. Costs related to adverse reactions and treatment administration were higher for glofitamab. With lower drug costs overall, glofitamab offers greater healthcare budget predictability compared with epcoritamab, which is expected to translate to cost savings at the broader healthcare system and population levels.