Our objective was to evaluate the recall and number needed to read (NNR) for the Cochrane RCT Classifier compared to and in combination with established search filters developed for Ovid MEDLINE and Embase.com. A gold standard set of 1,103 randomized controlled trials (RCTs) was created to calculate recall for the Cochrane RCT Classifier in Covidence, the Cochrane sensitivity-maximizing RCT filter in Ovid MEDLINE and the Cochrane Embase RCT filter for Embase.com. In addition, the classifier and the filters were validated in three case studies using reports from the Swedish Agency for Health Technology Assessment and Assessment of Social Services to assess impact on search results and NNR. The Cochrane RCT Classifier had the highest recall with 99.64% followed by the Cochrane sensitivity-maximizing RCT filter in Ovid MEDLINE with 98.73% and the Cochrane Embase RCT filter with 98.46%. However, the Cochrane RCT Classifier had a higher NNR than the RCT filters in all case studies. Combining the RCT filters with the Cochrane RCT Classifier reduced NNR compared to using the RCT filters alone while achieving a recall of 98.46% for the Ovid MEDLINE/RCT Classifier combination and 98.28% for the Embase/RCT Classifier combination. In conclusion, we found that the Cochrane RCT Classifier in Covidence has a higher recall than established search filters but also a higher NNR. Thus, using the Cochrane RCT Classifier instead of current state-of-the-art RCT filters would lead to an increased workload in the screening process. A viable option with a lower NNR than RCT filters, at the cost of a slight decrease in recall, is to combine the Cochrane RCT Classifier with RCT filters in database searches.

Managing high-variant product portfolios effectively is a crucial competitive advantage in offering mass customized products on saturated markets. Association Rule Mining (ARM) is a field of data mining determining frequent itemsets from historic transactions and deriving patterns of conclusion. This paper introduces a new approach to transfer ARM to feature-based configuration e.g. in the German automotive industry. Combined, existing apriori product knowledge is used in constraints to effectively lowering runtime by reducing the number of candidate-sets through introduction of a Boolean satisfiability check. For an efficient implementation, three different Apriori algorithms are tested and benchmarked on a generic dataset for different parameters. Results show a significant improvement in using SAT-based pre-screening while efficiency of the implementation depends on the given example.

With recent advancements in data-driven methods, there has been a growing interest in implementing AI in design. Despite this, a comprehensive understanding of the critical AI methods in design and how they support design practices remains lacking. To deepen our understanding, we conduct a comprehensive literature review and propose a novel, design-centric AI typology, associated with six AI assistance types for product service development. Our typology differs from traditional ones by shifting the focus from an algorithmic perspective to how models support design practice. Key findings highlight how these six design-centric AI methods support design practices in different ways, each with its own application challenges. Establishing a shared design-centric AI typology and assistance framework will enhance the understanding of how AI works differently and supports practitioners.

Active colloidal particles create flow around them due to non-equilibrium processes on their surfaces. In this paper, we infer the activity of such colloidal particles from the flow field created by them via deep learning. We first explain our method for one active particle, inferring the $2s$ mode (or the stresslet) and the $3t$ mode (or the source dipole) from the flow field data, along with the position and orientation of the particle. We then apply the method to a system of many active particles. We find excellent agreements between the predictions and the true values of activity. Our method presents a principled way to predict arbitrary activity from the flow field created by active particles.

Engineering design is inherently a collaborative process that requires active engagement and effective communication. Project-based Learning (PBL) is increasingly recognized for fostering these essential skills. However, instructors face challenges in objectively monitoring interactions and providing process-oriented feedback, particularly in large-scale settings where free-riders and disengaged participants affect team dynamics. This study introduces a generative AI approach to deliver real-time, scalable, and empathetic feedback that enhances team collaboration. Findings highlight the potential of AI-driven systems to improve student engagement and learning outcomes, though limitations remain in providing context-specific advice. A secure framework for AI integration in collaborative learning environments is also proposed.

Anatomical variations in the upper airway significantly impact the effectiveness of video laryngoscope blades. Existing literature on upper airway dynamics and blade design lacks a comprehensive framework to address these variations. The proposed model uses the extent of mouth opening with three demographic features and three anatomical features in the closed-mouth state to predict the anatomical features in the open-mouth state, which can support the design of a laryngoscope blade. Pearson’s correlation was studied to understand the correlation between the features, and the ordinary least square method was used to develop a model. For all three outputs, a separate model was developed, which gave R-squares of 0.98,0.74 and 0.94. The findings highlight the potential of data-driven approaches to optimize laryngoscope blade designs.

Traditional design automation enables parameterized customization but struggles with adapting to abstract or context-based user requirements. Recent advances in integrating large language models with script-driven CAD kernels provide a novel framework for context-sensitive, natural-language-driven design processes. Here, we present augmented design automation, enhancing parametric workflows with a semantic layer to interpret and execute functional, constructional, and effective user requests. Using CadQuery, experiments on a sandal model demonstrate the system’s capability to generate diverse and meaningful design variations from abstract prompts. This approach overcomes traditional limitations, enabling flexible and user-centric product development. Future research should focus on addressing complex assemblies and exploring generative design capabilities to expand the potential of this approach.

Lightweight design is critical for improving the efficiency and sustainability of engineering applications. Laminated composites, with their high strength-to-weight ratio and tailored material properties, play a key role but introduce interlaminar stresses, particularly near free edges where delamination failures often occur. Addressing these stresses typically requires computationally expensive 3D finite element simulations, limiting their use in early design stages. This study presents a machine learning approach using Gaussian process regression and artificial neural networks to efficiently predict interlaminar stresses based on in-plane stress data from shell FE simulations. Achieving high predictive accuracy, this method enables cost-effective, early-stage composite design optimization under complex loading scenarios.

Effective product development relies on creating a requirements document that defines the product’s technical specifications, yet traditional methods are labor-intensive and depend heavily on expert input. Large language models (LLMs) offer the potential for automation but struggle with limitations in prompt engineering and contextual sensitivity. To overcome these challenges, we developed ReqGPT, a domain-specific LLM fine-tuned on Mistral-7B-Instruct-v0.2 using 107 curated requirements lists. ReqGPT employs a standardized prompt to generate high-quality documents and demonstrated superior performance over GPT-4 and Mistral in multiple criteria based on ISO 29148. Our results underscore ReqGPT’s efficiency, accuracy, cost-effectiveness, and alignment with industry standards, making it an ideal choice for localized use and safeguarding data privacy in technical product development.

Engineering of lightweight and robust structures is significant in mechanical engineering. Nevertheless, weight optimization of such structures leads to undesirable vibrations. Modal analysis is a common technique used in industry to investigate vibration behaviour. The classification of the mode shapes resulting from the analysis is conducted through human visual inspection, which can be time-consuming and susceptible to error. This paper presents an exploratory study investigating the potential of ML methods to classify three-dimensional vibration modes of truck frame structures. The aim is to evaluate the potential of such an approach to automate the modal analysis process to streamline the development process. As a result, the developed ML model can classify the vibration modes with high performance and additionally demonstrates flexibility regarding changes in geometry topology.

Recent advancements in machine learning (ML) offer substantial potential for enhancing product development. However, adoption in companies remains limited due to challenges in framing domain-specific problems as ML tasks and selecting suitable ML algorithms, requiring expertise often lacking. This study investigates the use of large language models (LLMs) as recommender systems for facilitating ML implementation. Using a dataset derived from peer-reviewed publications, the LLMs were evaluated for their ability to recommend ML algorithms for product development-related problems. The results indicate moderate success, with GPT-4o achieving the highest accuracy by recommending suitable ML algorithms in 61% of cases. Key limitations include inaccurate recommendations and challenges in identifying multiple sub-problems. Future research will explore prompt engineering to improve performance.

Visual-Language (VL) models offer potential for advancing Engineering Design (ED) by integrating text and visuals from technical documents. We review VL applications across ED phases, highlighting three key challenges: (i) understanding how functional and structural information is complementarily expressed by text and images, (ii) creating large-scale multimodal design datasets and (iii) improving VL models’ ability to represent ED knowledge. A dataset of 1.5 million text-image pairs and an evaluation dataset for cross-modal information retrieval were developed using patents. By Fine-tuning and testing the CLIP base model on these datasets, we identified significant limitations in VL models’ capacity to capture fine-grained technical details required for precision-driven ED tasks. Based on these findings, we propose future research directions to advance VL models for ED applications.

Natural Language Processing (NLP) has been widely applied in design, particularly for analyzing technical documents like patents and scientific papers to extract engineering design knowledge. This work aims to enhance this process by integrating the Axiomatic Design methodology with NLP techniques applied to patent texts. The objectives are to (1) extract Functional requirements (FRs) and Design parameters (DPs), and (2) identify how FRs and DPs are related in text (Axiomatic relations). The second objective is particularly challenging due to limited focus on understanding semantic relations in literature, and previous studies often extract Axiomatic relations in an unstructured way. The approach achieves 60% precision for the first objective and 30-50% for the second. Moreover, a case study shows the practical application of this methodology to assist the work of designers.

Generative artificial intelligence (GenAI) has the potential to further revolutionize Computer-Aided Design (CAD) by recognizing patterns, making predictions, and generating automated design suggestions. This paper presents a systematic literature review that examines the current state of research on the use of GenAI in CAD-based product development. With a focus on 3D modelling, it provides an overview of current approaches, most used datasets and commonly used AI models. Four application areas where GenAI can enhance CAD were derived: Design generation, Design reconstruction, Design retrieval, and Design modification. In total, 47 papers were selected, analysed and categorised.

Bolted joints are critical for maintaining structural integrity and reliability. Accurate prediction of parameters is essential for optimal performance. Traditional methods often fail to capture the non-linear behavior or require significant computational resources, limiting accuracy and efficiency. This study addresses these limitations by combining empirical data with a feed-forward neural network. Leveraging experimental data and systematic preprocessing, the model effectively captures nonlinear relationships, including rescaling output variables to address scale discrepancies, achieving 95% predictive accuracy. While limited dataset size restricts generalizability, the findings demonstrate the potential of neural networks as a reliable, efficient alternative for bolted joint design. Future work aims to expand datasets and explore hybrid modeling techniques to enhance applicability.

This paper investigates the effectiveness of machine learning models in predicting customer-relevant functional attributes of vehicles based on selected design variables, using a limited automobile market dataset. By comparing machine learning algorithms such as Support Vector Regression, k-Nearest Neighbour Regression, and Lasso Regression, the study evaluates the models’ predictive accuracy and their potential application in automotive design. The findings highlight both the opportunities and limitations of these methods, emphasising their capacity to support data-driven decision-making despite constraints posed by dataset size, as encountered in real-world, early-stage automotive platform strategies.

Functional decomposition (FD) is essential for simplifying complex systems in engineering design but remains a resource-intensive task reliant on expert knowledge. Despite advances in artificial intelligence, the automation of FD remains underexplored. This study introduces the use of GPT-4o, enhanced with a proposed Monte Carlo tree search for functional decomposition (MCTS-FD) algorithm, to automate FD. The approach is evaluated qualitatively by comparing outputs with those of graduate engineering students and quantitatively by assessing metrics such as structural integrity and semantic accuracy. The results show that GPT-4o, enhanced by MCTS-FD, outperforms smaller models in error rates and graph connectivity, highlighting the potential of large language models to automate FD with human-like accuracy.

This study presents a methodology for leveraging an LLM to generate user-centered recommendations in design for sustainable behavior. A survey of 50 users captured reasonings for evaluating thermostats’ eco-friendliness and sustainable design features. Through in-context learning, GPT-4o learned to take user perspectives for similar evaluations. The model classified 196 thermostats by eco-friendliness and design intervention types—persuasive, decisive, or both. Analysis of user sentiment and ratings of these thermostats’ reviews showed persuasive designs, which offer users behavioral control, received higher satisfaction. GPT-4o extracted features from these classifications to generate design recommendations. This method is a scalable approach for identifying user preferences and informing sustainable design decisions.