Task Analysis Research Articles

A Wizard of Oz Study of Guidance Strategies and Dynamics.

Co-adaptive guidance in visual analytics is a mixed-initiative process in which both the user and the system work together to support each other in solving a given analysis task. While previous studies show the effectiveness of guidance, the impact of guidance design decisions (e.g., the suggestions' timing, contextualization, or adaptation) and misguidance often remain under-investigated. To investigate these aspects and examine a variety of guidance interaction patterns in a realistic analysis scenario, we present a Wizard of Oz (WOz) study setup in which pairs of participants take the user's and the system's roles, respectively. As users perform their analysis tasks, they are observed by wizards who provide just-in-time guidance as they see fit. Moreover, we designed the study so that wizards would occasionally and unknowingly provide misguidance during the analysis to investigate the users' confidence in guidance systems. We recruited two groups of participants (12 wizards and 12 users) and paired each participant with two from the other group, obtaining 48 observations. We report insights on interactions between users and wizards. By analyzing these interaction dynamics and the guidance strategies the wizards apply, we derive recommendations for implementing and evaluating future co-adaptive guidance systems.

XGBoost-Enhanced Graph Neural Networks: A New Architecture for Heterogeneous Tabular Data

Graph neural networks (GNNs) perform well in text analysis tasks. Their unique structure allows them to capture complex patterns and dependencies in text, making them ideal for processing natural language tasks. At the same time, XGBoost (version 1.6.2.) outperforms other machine learning methods on heterogeneous tabular data. However, traditional graph neural networks mainly study isomorphic and sparse data features. Therefore, when dealing with tabular data, traditional graph neural networks encounter challenges such as data structure mismatch, feature selection, and processing difficulties. To solve these problems, we propose a novel architecture, XGNN, which combines the advantages of XGBoost and GNNs to deal with heterogeneous features and graph structures. In this paper, we use GAT for our graph neural network model. We can train XGBoost and GNN end-to-end to fit and adjust the new tree in XGBoost based on the gradient information from the GNN. Extensive experiments on node prediction and node classification tasks demonstrate that the performance of our proposed new model is significantly improved for both prediction and classification tasks and performs particularly well on heterogeneous tabular data.

Application of machine reading comprehension techniques for named entity recognition in materials science

Materials science is an interdisciplinary field that studies the properties, structures, and behaviors of different materials. A large amount of scientific literature contains rich knowledge in the field of materials science, but manually analyzing these papers to find material-related data is a daunting task. In information processing, named entity recognition (NER) plays a crucial role as it can automatically extract entities in the field of materials science, which have significant value in tasks such as building knowledge graphs. The typically used sequence labeling methods for traditional named entity recognition in material science (MatNER) tasks often fail to fully utilize the semantic information in the dataset and cannot effectively extract nested entities. Herein, we proposed to convert the sequence labeling task into a machine reading comprehension (MRC) task. MRC method effectively can solve the challenge of extracting multiple overlapping entities by transforming it into the form of answering multiple independent questions. Moreover, the MRC framework allows for a more comprehensive understanding of the contextual information and semantic relationships within materials science literature, by integrating prior knowledge from queries. State-of-the-art (SOTA) performance was achieved on the Matscholar, BC4CHEMD, NLMChem, SOFC, and SOFC-Slot datasets, with F1-scores of 89.64%, 94.30%, 85.89%, 85.95%, and 71.73%, respectively in MRC approach. By effectively utilizing semantic information and extracting nested entities, this approach holds great significance for knowledge extraction and data analysis in the field of materials science, and thus accelerating the development of material science.Scientific contributionWe have developed an innovative NER method that enhances the efficiency and accuracy of automatic entity extraction in the field of materials science by transforming the sequence labeling task into a MRC task, this approach provides robust support for constructing knowledge graphs and other data analysis tasks.

Human error identification and risk assessment in loading and unloading of petroleum products by road trucks using the SHERPA and fuzzy inference system method

Human error constitutes one of the primary causes of accidents, particularly in the context of loading and unloading operations involving road trucks, especially those carrying petroleum products. The process of identifying and evaluating human errors within these operations involves several key steps. Initially, all sub-tasks associated with loading and unloading are meticulously identified and analyzed utilizing Hierarchical Task Analysis (HTA), achieved through direct observation, document examination, and interviews. Subsequently, potential human error modes within each task are delineated using the Systematic Human Error Reduction and Prediction Approach (SHERPA). Finally, essential data for determining the criticality, probability, and severity of each error are gathered through expert elicitation and the application of Fuzzy Inference Systems (FIS).Through the analysis of SHERPA worksheets, a total of 37 errors during loading operations and 14 errors during unloading operations of petroleum products were identified. Among these errors, the predominant category during loading operations was action errors, comprising 31 instances, while communication errors were the least frequent, occurring only twice. Similarly, action errors were most prevalent during unloading operations, constituting 13 instances. These errors were further categorized and ranked based on their risk levels, resulting in 27 levels for loading operations and 12 levels for unloading operations.The consistent occurrence of action errors underscores the need for implementing control measures to mitigate their frequency and severity. Such strategies may include periodic training sessions to reinforce proper work procedures and the development of monitoring checklists, among other interventions.

Rigorous development and comparison of multi-dimensional reactor models encompassing the catalyst domains for steam methane reforming

Hydrogen is being considered as a possible large-scale carbon-free industrial combustion and transportation fuel. Steam methane reforming (SMR) reaction is the most dominant industrial hydrogen production method. This paper develops comprehensive models and delves into the influences of the multi-dimensional reactor and catalyst domains in SMR. Several single-scale or multi-scale SMR packed-bed reactor models are developed with the Aspen Custom Modeler and solved by the built-in finite difference method. These models are categorized into pseudo-homogeneous and heterogeneous models. This study thoroughly compares and discusses the effects of axial and/or radial mixing due to mass/heat dispersion as well as interfacial and intraparticle mass/heat transport phenomena across various types of reactor-catalyst configurations. Furthermore, insights into the prediction accuracy of each model under various operating conditions are provided. This research fills a gap in the existing literature by developing rigorous dynamic models for SMR reactors to investigate their profiles and performances, offering a clearer understanding of the relative importance of different modeling approaches in predicting the behavior of SMR reactors. The findings provide valuable guidelines for researchers and industry professionals in selecting the appropriate model for their design, analysis, optimization, and control tasks.

Video Coding for Machines: Compact Visual Representation Compression for Intelligent Collaborative Analytics.

As an emerging research practice leveraging recent advanced AI techniques, e.g. deep models based prediction and generation, Video Coding for Machines (VCM) is committed to bridging to an extent separate research tracks of video/image compression and feature compression, and attempts to optimize compactness and efficiency jointly from a unified perspective of high accuracy machine vision and full fidelity human vision. With the rapid advances of deep feature representation and visual data compression in mind, in this paper, we summarize VCM methodology and philosophy based on existing academia and industrial efforts. The development of VCM follows a general rate-distortion optimization, and the categorization of key modules or techniques is established including feature-assisted coding, scalable coding, intermediate feature compression/optimization, and machine vision targeted codec, from broader perspectives of vision tasks, analytics resources, etc. From previous works, it is demonstrated that, although existing works attempt to reveal the nature of scalable representation in bits when dealing with machine and human vision tasks, there remains a rare study in the generality of low bit rate representation, and accordingly how to support a variety of visual analytic tasks. Therefore, we investigate a novel visual information compression for the analytics taxonomy problem to strengthen the capability of compact visual representations extracted from multiple tasks for visual analytics. A new perspective of task relationships versus compression is revisited. By keeping in mind the transferability among different machine vision tasks (e.g. high-level semantic and mid-level geometry-related), we aim to support multiple tasks jointly at low bit rates. In particular, to narrow the dimensionality gap between neural network generated features extracted from pixels and a variety of machine vision features/labels (e.g. scene class, segmentation labels), a codebook hyperprior is designed to compress the neural network-generated features. As demonstrated in our experiments, this new hyperprior model is expected to improve feature compression efficiency by estimating the signal entropy more accurately, which enables further investigation of the granularity of abstracting compact features among different tasks.

A fused convolutional transformer for voltammetric electronic tongue analysis tasks

Electronic tongue and artificial intelligence have been evaluated as new rapid analysis technologies in complex solution research. The use of nanomaterials to modify the screen-printed carbon electrode (SPCE) increased the specific surface area and electrocatalytic active sites of the electrode. The voltammetric electronic tongue utilizes multiple cross sensitive sensors to describe the overall characteristics of complex samples, and the complementary information between sensors provides rich and comprehensive descriptions for analyzing various components. The use of electronic tongue for rapid analysis of water quality has gradually become a trend, but the overlapping phenomenon of peaks formed by the similar redox potential of each component in water and the interference of the substrate makes it difficult to obtain effective information of each component in water quality. Here, we propose a Fused Convolutional Transformer model (FCvT) that removes the limitation of overlapping peaks by fusing local features and global complementary features to quantify components in complex solutions. The performance of FCvT was evaluated by testing synthetic water samples consisting of four standardized substance solutions and compared with partial least squares (PLS), support vector regression (SVR), artificial neural networks (ANN), and convolutional neural networks (CNN). The results show that the FCvT obtains the lowest mean absolute percentage error in the set of independent tests.The average quantization error of FCvT is reduced by 34.4 % relative to other methods.

Screening COVID-19 from chest X-ray images by an optical diffractive neural network with the optimized F number

The COVID-19 pandemic continues to significantly impact people’s lives worldwide, emphasizing the critical need for effective detection methods. Many existing deep learning-based approaches for COVID-19 detection offer high accuracy but demand substantial computing resources, time, and energy. In this study, we introduce an optical diffractive neural network (ODNN-COVID), which is characterized by low power consumption, efficient parallelization, and fast computing speed for COVID-19 detection. In addition, we explore how the physical parameters of ODNN-COVID affect its diagnostic performance. We identify the F number as a key parameter for evaluating the overall detection capabilities. Through an assessment of the connectivity of the diffractive network, we established an optimized range of F number, offering guidance for constructing optical diffractive neural networks. In the numerical simulations, a three-layer system achieves an impressive overall accuracy of 92.64% and 88.89% in binary- and three-classification diagnostic tasks. For a single-layer system, the simulation accuracy of 84.17% and the experimental accuracy of 80.83% can be obtained with the same configuration for the binary-classification task, and the simulation accuracy is 80.19% and the experimental accuracy is 74.44% for the three-classification task. Both simulations and experiments validate that the proposed optical diffractive neural network serves as a passive optical processor for effective COVID-19 diagnosis, featuring low power consumption, high parallelization, and fast computing capabilities. Furthermore, ODNN-COVID exhibits versatility, making it adaptable to various image analysis and object classification tasks related to medical fields owing to its general architecture.

Game provenance graph-based representation learning vs metrics-based machine learning: An empirical comparison on predictive game analytics tasks

Predicting events and behavior in games is a crucial task of game analytics, both during gameplay, such as predicting wins, and after, like forecasting player churn. Game metrics have been widely adopted due to their human-readable and task-oriented nature. However, these metrics vary between games and require expert knowledge. Game analytics has evolved with techniques that primarily handle non-tabular and sequential data, like graphs and time series. However, existing methods still depend on traditional game metrics from extensive feature engineering. In parallel, graph representation learning has been applied to game provenance graphs — structures that log game events as nodes and link them through causal relationships. This approach may allow analysts to bypass traditional game metrics and learn relevant downstream features during training. This paper compares this metric-less graph-based representation learning with traditional metric-based machine learning in two predictive game analytics tasks: win prediction and hit prediction. Our results show that the same graph-based representation learning architecture achieves competitive results in win prediction and outperforms metric-based approaches in hit prediction, suggesting its suitability for predictive game analytics tasks. This method could significantly benefit as it does not require game-specific metric expertise, and its learned features are versatile across multiple tasks.

Global sensitivity analysis with multifidelity Monte Carlo and polynomial chaos expansion for vascular haemodynamics.

Computational models of the cardiovascular system are increasingly used for the diagnosis, treatment, and prevention of cardiovascular disease. Before being used for translational applications, the predictive abilities of these models need to be thoroughly demonstrated through verification, validation, and uncertainty quantification. When results depend on multiple uncertain inputs, sensitivity analysis is typically the first step required to separate relevant from unimportant inputs, and is key to determine an initial reduction on the problem dimensionality that will significantly affect the cost of all downstream analysis tasks. For computationally expensive models with numerous uncertain inputs, sample-based sensitivity analysis may become impractical due to the substantial number of model evaluations it typically necessitates. To overcome this limitation, we consider recently proposed Multifidelity Monte Carlo estimators for Sobol' sensitivity indices, and demonstrate their applicability to an idealized model of the common carotid artery. Variance reduction is achieved combining a small number of three-dimensional fluid-structure interaction simulations with affordable one- and zero-dimensional reduced-order models. These multifidelity Monte Carlo estimators are compared with traditional Monte Carlo and polynomial chaos expansion estimates. Specifically, we show consistent sensitivity ranks for both bi- (1D/0D) and tri-fidelity (3D/1D/0D) estimators, and superior variance reduction compared to traditional single-fidelity Monte Carlo estimators for the same computational budget. As the computational burden of Monte Carlo estimators for Sobol' indices is significantly affected by the problem dimensionality, polynomial chaos expansion is found to have lower computational cost for idealized models with smooth stochastic response.

Visual Interactive Parameter Selection for Temporal Blind Source Separation

Many fields of science and industry collect and analyze multivariate time-varying measurements, e.g., healthcare, geophysics, or finance. Such data is often high-dimensional, correlated, and noisy. Experts are interested in latent components of the dataset, but due to the aforementioned properties these are difficult to obtain. Temporal Blind Source Separation (TBSS) is a suitable and well-established framework for these data. However, the large choice of methods and their tuning parameters impedes the effective use of TBSS in practice. The goal of Visual Analytics (VA) is to create powerful analytic tools by combining the strengths of humans and computers. We designed, developed, and evaluated VA contributions in previous work to support TBSS-related analysis tasks. In this paper, we highlight the benefits and opportunities of VA concepts for statistic-oriented problems using a real-world TBSS application example with a dataset of climate and meteorological measurements in Italy.

Specifics of Due Diligence procedure in mergers and acquisitions in the Russian Federation

Aim. To study the elements and main features of the Due Diligence procedure, including mandatory and additional stages of due diligence of Russian companies, as well as taking into account the peculiarities of the Russian legislation and regulatory bodies.Objectives. To perform an in-depth theoretical analysis of the main tasks of the Due Diligence procedure in M&A of enterprises; to develop proposals for improving the methodology of the Due Diligence procedure in the Russian Federation (RF).Methods. The authors used general scientific research methods, including data collection and processing, content analysis, generalization, systematization, analysis of legislation and legal practice in the field of mergers and acquisitions, as well as the experience of Russian enterprises in conducting the Due Diligence procedure.Results. The Due Diligence procedure is an important stage of corporate transactions in the M&A process in Russia. The article shows that one of the key features is the necessity to take into account specific legal and tax aspects of the Russian legislation when conducting due diligence. It is necessary to take into account not only the risks, but also the influence of state regulators and regulatory bodies on the results of the transaction. Based on the results of the analysis and systematization of the information obtained, general methods and recommendations in the course of the Due Diligence procedure were determined. The statistical data on M&A transactions in Russia are considered, the attention is emphasized on the importance of Due Diligence at the rapid growth of M&A transactions. It is proved that the Due Diligence procedure increases the transparency of Russian companies and contributes to the improvement of the investment climate in the country.Conclusions. The study of the peculiarities of Due Diligence in M&A in Russia allows us to focus on the importance of legal analysis, financial and economic evaluation, as well as personnel and corporate culture assessment. Legal analysis provides an opportunity to identify and assess legal risks and problems related to the company's operations. Financial and economic analysis and audit help to identify and assess financial risks and problems, current and potential. Personnel and corporate culture assessment helps to understand how well they are aligned with the business objectives and strategy of the transaction. All three areas of assessment must comply with the requirements of domestic legislation and take into account the specifics of the labor market in Russia. A proper Due Diligence assessment of all three areas will help to detect possible risks and problems related to legal, financial and personal aspects of the company's activities, as well as provide a reliable basis for making correct and informed decisions regarding the M&A transaction.

The quality of prospective mathematics teachers' dynamic geometry tasks in terms of the coordination between mathematical depth levels and technological actions

AbstractBackgroundTeachers must have the skills to find, select, design and use technology‐based mathematical activities that focus on high‐level cognitive demands—supporting student reasoning. However, they experience various difficulties in creating, planning and making decisions about how and when to utilize technological affordances.ObjectivesThis study aims to examine the quality of prospective school mathematics teachers' (PMTs) tasks containing the use of dynamic geometry software through the analysis of the coordination between mathematical depth levels (MDLs) and technological actions (TAs), according to Trocki and Hollebrands', Digital Experiences in Mathematics Education, 2018, 4, 110–138, Dynamic Geometry Task Analysis Framework.MethodsWithin groups, PMTs designed lesson plans that included dynamic geometry tasks (DGTs) based on the learning outcomes of the middle school mathematics curriculum. They then created micro‐teachings of these tasks. The participants' DGTs and transcripts of their microteaching videos were the main data sources.ResultsWe found that many prompts included no TA or a single TA, regardless of the prompts' MDLs. Moreover, the results showed that the high‐depth prompts aimed to contribute to students' reasoning processes on geometric concepts through the TAs, which required specific tools such as dragging and sliding to get generalizations. Although the groups differed in the quality and number of DGTs, the groups' lesson plans mostly contained low‐ or medium‐quality DGTs regarding the coordination between MDLs and TAs.ConclusionsInforming PMTs about high‐quality technology tasks and increasing their awareness on this matter is a priority to encourage them to incorporate innovative technological activities in their mathematics instruction.

Protocols versus practice: unravelling clinical checking variations in community pharmacies in England-a multi-method study.

Standardisation, a widely accepted concept for risk management, entails designing and implementing task-specific operating procedures. In community pharmacies, Standardised Operating Procedures (SOPs) are a mandatory requirement and are recognised as essential for upholding safety and quality. This study aimed to investigate community pharmacists' (CPs) compliance with SOPs when checking prescriptions, and the reasons for variations between standardised protocols and practice. Eight sets of SOPs underwent hierarchical task analysis (HTA) to generate a normative description of clinical checking execution as per protocols. Subsequently, twelve CPs were engaged in a simulated clinical checking exercise, verbalising their thoughts while checking virtual prescriptions. Transcribed data underwent content analysis, aligned with a descriptive model to uncover engagement patterns, and disparities between SOPs and CPs' practices. Finally, a focus group discussion took place to contextualise the observed variations. HTA aided in constructing a clinical checking model with six primary subtasks and 28 lower subtasks. CPs often omitted subtasks during checks, diverging from prescribed protocols. These deviations, observed in controlled environment, reveal an ingrained aspect within the professional culture of pharmacists, where there may be a tendency not to strictly adhere to protocols, despite variations in work conditions. Contributing factors to this culture include the exercise of professional judgment, reliance on others, and prioritisation of patient preferences. This study highlights ongoing deviations from SOPs during clinical prescription checks in community pharmacies, suggesting a cultural tendency. Future research should delve into risk management strategies for these deviations and address the delicate balance between flexibility and stringent compliance.

Application of a simple two-point measurement strategy with emphasizing low-frequency signals to the force monitoring of cables

When dealing with vibration tests of tensioned suspenders and cables, signals containing relatively high frequency components in comparing to the fundamental frequency are normally required in order to carry out reliable diagnoses for force estimations. Meanwhile, the necessity of obtaining a precise and broadband spectrum of the vibration signals would inevitably need to acquire relatively detailed time responses. One typical difficulty, which arises when incorporating vibration analyses into monitoring tasks, is the inefficiency caused by the post- processing for the huge amounts of time data. In this work, a set of simple yet effective formulas associated with two-point measurements will be briefly addressed. The proposed formulas are derived based on the continuous formulations of the string model, and the corresponding applications to the tension estimations of slender members will be demonstrated using numerical and experimental data related to stayed cables of various restrained scenarios and linked-suspenders as well. The purposes of this work are twofold; firstly to introduce the potential usefulness of the proposed measurement strategy to form effective schemes for the tension assessments; secondly to evolve an efficient incorporation of wireless measuring techniques without the need of processing cumbersome time data which are the common consequences the analysts would encounter in the analysis tasks. Finally, this work aims to highlight practical ideas in guiding the use of the proposed techniques in the monitoring of cables/suspenders in field.

DualGCN: Exploring Syntactic and Semantic Information for Aspect-Based Sentiment Analysis.

The task of aspect-based sentiment analysis aims to identify sentiment polarities of given aspects in a sentence. Recent advances have demonstrated the advantage of incorporating the syntactic dependency structure with graph convolutional networks (GCNs). However, their performance of these GCN-based methods largely depends on the dependency parsers, which would produce diverse parsing results for a sentence. In this article, we propose a dual GCN (DualGCN) that jointly considers the syntax structures and semantic correlations. Our DualGCN model mainly comprises four modules: 1) SynGCN: instead of explicitly encoding syntactic structure, the SynGCN module uses the dependency probability matrix as a graph structure to implicitly integrate the syntactic information; 2) SemGCN: we design the SemGCN module with multihead attention to enhance the performance of the syntactic structure with the semantic information; 3) Regularizers: we propose orthogonal and differential regularizers to precisely capture semantic correlations between words by constraining attention scores in the SemGCN module; and 4) Mutual BiAffine: we use the BiAffine module to bridge relevant information between the SynGCN and SemGCN modules. Extensive experiments are conducted compared with up-to-date pretrained language encoders on two groups of datasets, one including Restaurant14, Laptop14, and Twitter and the other including Restaurant15 and Restaurant16. The experimental results demonstrate that the parsing results of various dependency parsers affect their performance of the GCN-based models. Our DualGCN model achieves superior performance compared with the state-of-the-art approaches. The source code and preprocessed datasets are provided and publicly available on GitHub (see https://github.com/CCChenhao997/DualGCN-ABSA).

Measuring Operational Efficiency: A Functional Analysis of Core and Non-Core Activities in a Leading Romanian Oil and Gas Retail Company

Abstract This paper employs a qualitative approach to conduct functional analysis within one of the main players in the Romanian market in the retail oil and gas industry, utilizing the company's internal intranet platform to gather data. The study unfolds in distinct stages, beginning with department selection, followed by task breakdown, activity categorization, data collection, analysis, and insight sharing. The analysis reveals a comprehensive view of the company's operational landscape, guiding strategic decisions and resource optimization. Key conclusions highlight the significance of core activities for revenue and growth, the value of diverse employee typologies, and opportunities for efficiency enhancement. The study's insights offer practical directions for managing core competencies, fostering informed collaboration, and adapting to industry changes. Future recommendations include leveraging this analysis for future strategy, investment, and innovation.

Integrated Language and Science & Technology Instruction: A Cognitive Task Analysis of the Required Teacher Expertise

ABSTRACT Integrated language arts and science & technology (ILS&T) instruction can make learning more meaningful and improve student learning outcomes for both subjects. Although the literature has stressed that such integrated instruction presents pedagogical challenges for elementary school teachers, little is yet known about the specific characteristics of the required pedagogical repertoire. To gain a better understanding of the teacher knowledge and skills required for ILS&T instruction, a cognitive task analysis was conducted. This analysis resulted in a hierarchy consisting of the various required constituent teacher skills for ILS&T instruction. The cognitive task analysis also revealed what knowledge teachers rely on and identified the factors influencing the complexity of ILS&T instruction. The findings presented here contribute to our knowledge about what ILS&T instruction requires from teachers and can serve as a basis for the development of substantiated, 4C/ID-based teacher professional development trajectories. Furthermore, it provides valuable insight into the pros and cons of a relatively new, systematic way of analyzing teacher expertise.

Negation handling for sentiment analysis task: approaches and performance analysis

Negation plays an essential role in sentiment analysis within natural language processing (NLP). Its integration involves two key aspects: identifying the scope of negation and incorporating this information into the sentiment model. Before delving into scope detection, the specific negation cue must be identified, with explicit and implicit negation cues being the two main types. Various methodologies, such as rule-based, machine learning, and hybrid approaches, address the negation scope detection challenge. Strategies for leveraging negation information in sentiment models encompass heuristic polarity modification, feature space augmentation, end-to-end approach, and hierarchical multi-task learning. Notably, there is a need for more studies addressing implicit negation cue detection, even within the state-of-the-art bidirectional encoder representation for transformers (BERT) approach. Some studies have employed reinforcement learning and hybrid techniques to address the implicit negation problem. Further exploration, particularly through a hybrid and multi-task learning approach, is warranted to make potential contributions to the nuanced challenges of handling negation in sentiment analysis, especially in complex sentence structures.

Pre-ANDA strategy and Human Factors activities to de-risk pharmaceutical companies ANDA submission of drug–device combination products: case study of a formative Comparative Use Human Factors study

ABSTRACT Background This article presents a strategy that a Drug Delivery Device Developer (DDDD) has adopted to support Abbreviated New Drug Application (ANDA) submissions of drug–device combination products. As per the related FDA guidance, a threshold analysis should be compiled. If ‘other differences’ between the Reference Listed Drug (RLD) and the generic drug devices are identified, a Comparative Use Human Factors (CUHF) study may be requested. Methods The DDDD performed task analysis and physical comparison to assess the pen injector design differences. Then, a formative CUHF study with 25 participants simulating injections using both RLD and the generic pen injectors was conducted. Results After each participant completed four simulated injections, similar type and rates of use error between the RLD (0.70) and generic (0.68) pen injectors were observed. Conclusion DDDDs can support pharmaceutical companies in the ANDA submission strategy of their drug–device combination product by initiating comparative task analysis and physical comparison of the device as inputs for the threshold analysis. If ‘other differences’ are identified, a formative CUHF study can be performed. As shown in our case study, this approach can be leveraged to support the sample size calculation and non-inferiority margin determination for a CUHF study with the final combination product.