Single Artificial Intelligence Research Articles

Arctic tern-optimized weighted feature regression system for predicting bridge scour depth

This paper presents a pioneering artificial intelligence (AI) solution – the Arctic Tern-Optimized Weighted Feature Least Squares Support Vector Regression (ATO-WFLSSVR) system to aid civil engineers in accurately predicting scour depth at bridges. This prediction system amalgamates the strengths of hybrid models by uniting a metaheuristic optimization algorithm with weighted features and least squares support vector regression (WFLSSVR). The metaheuristic algorithm concurrently optimizes all hyperparameters of constituent WFLSSVR models, resulting in a highly effective system. Validation involves a comprehensive assessment using two case studies, which include datasets of scour depths across various complexities and pier foundation types. Comparative analyses against single AI models, conventional ensemble models, hybrid techniques, and empirical methods demonstrate that ATO-WFLSSVR's reliability outperforms others in performance evaluation metrics. Specifically, for the field dataset, ATO-WFLSSVR achieves MAPE and R values of 20.92% and 0.9435, respectively, and for scour depth data at complex pier foundations, it records MAPE and R values of 6.49% and 0.9384, respectively. The automated predictive analytics underscore the robustness, efficiency, and stability of ATO-WFLSSVR compared to existing methods. This study's notable contributions include the development of an innovative optimization algorithm named Arctic Terns Optimizer (ATO), proficiency in solving high-dimensional optimization problems, and the creation of a user-friendly graphical interface system, a promising tool for civil engineers to estimate scour depth at bridges. Further testing and evaluation of ATO-WFLSSVR across diverse datasets encompassing more complex scenarios are recommended. The data and source code for this study are currently accessible at https://www.researchgate.net/profile/Jui-Sheng-Chou/publications.

A hybrid artificial intelligence approach for modeling the carbonation depth of sustainable concrete containing fly ash

One of the major challenges in the civil engineering sector is the durability of reinforced concrete structures against carbonation during the physico-chemical process of interaction of hydrated cementitious composites with carbon dioxide. This aggressive process causes carbon penetration into the reinforcement part, which affects the behavior of the structure during its lifetime due to corrosion risk. A countermeasure is using alternative cementitious materials to improve concrete texture and resist increased carbonation depth (CD). Considering that the CD test requires a long time and a skilled technician, this study strives to provide an alternative approach by moving from traditional laboratory-based methods towards artificial intelligence (AI) techniques for modeling the CD of sustainable concrete containing fly ash (CCFA). Despite the development of single AI models so far, it is undeniable that utilizing metaheuristic optimization techniques in the form of hybrid models can improve their performance. To this end, a new hybrid model from the integration of biogeography-based optimization (BBO) technique with artificial neural network (ANN) is developed for the first time to estimate the CD of CCFA. The error distribution results revealed that 59% of the ANN predictions had errors within the range of (− 1 mm, 1 mm], while the corresponding percentage for the ANN-BBO predictions was 70%, indicating an 11% reduction in the prediction errors by the proposed hybrid model. Furthermore, A10-index highlighted a performance improvement of 78% for the hybrid model, which met the closeness of the predicted values to the observed ones, so that the value of this index for models of ANN and ANN-BBO was 0.5019 and 0.8947, respectively. Analyzing the cross-validation confirmed the reliability and generalizability of the developed model. Also, the three most influential variables in estimating the CD were exposure time (27%), carbon dioxide concentration (22%), and water/binder (18%), respectively. Finally, the superiority of the ANN-BBO model was verified by comparing it with previous studies’ models.

Fusion-based approach for hydrometeorological drought modeling: a regional investigation for Iran.

The objective of this study was to model a new drought index called the Fusion-based Hydrological Meteorological Drought Index (FHMDI) to simultaneously monitor hydrological and meteorological drought. Aiming to estimate drought more accurately, local measurements were classified into various clusters using the AGNES clustering algorithm. Four single artificial intelligence (SAI) models-namely, Gaussian Process Regression (GPR), Ensemble, Feedforward Neural Networks (FNN), and Support Vector Regression (SVR)-were developed for each cluster. To promote the results of single of products and models, four fusion-based approaches, namely, Wavelet-Based (WB), Weighted Majority Voting (WMV), Extended Kalman Filter (EKF), and Entropy Weight (EW) methods, were used to estimate FHMDI in different time scales, precipitation, and runoff. The performance of single and combined products and models was assessed through statistical error metrics, such as Kling-Gupta efficiency (KGE), Mean Bias Error (MBE), and Normalized Root Mean Square Error (NRMSE). The performance of the proposed methodology was tested over 24 main river basins in Iran. The validation results of the FHMDI (the compliance of the index with the pre-existing drought index) revealed that it accurately identified drought conditions. The results indicated that individual products performed well in some river basins, while fusion-based models improved dataset accuracy more compared to local measurements. The WMV with the highest accuracy (lowest NRMSE) had a good performance in 60% of the cases compared to all other products and fusion-based models. WMV also showed higher efficiency in 100% of the cases than all other fusion-based and SAI models for simultaneous hydrological and meteorological drought estimation. In light of these findings, we recommend the use of fusion-based approach to improve drought modeling.

The Promise and Pitfalls of AI-Generated Anatomical Images: Evaluating Midjourney for Aesthetic Surgery Applications.

The rapid advancement of generative artificial intelligence (AI) systems, such as Midjourney, has paved the way for their use in medical training, producing computer-generated images. However, despite clear disclosures stating that these images are not intended for medical consultations, their accuracy and realism are yet to be thoroughly examined. A series of requests were addressed to the Midjourney AI tool, a renowned generative artificial intelligence application, with a focus on depicting appropriate systemic anatomy and representing aesthetic surgery operations. Subsequently, a blinded panel of four experts, with years of experience in anatomy and aesthetic surgery, assessed the images based on three parameters: accuracy, anatomical correctness, and visual impact. Each parameter was scored on a scale of 1-5. All of images produced by Midjourney exhibited significant inaccuracies and lacked correct anatomical representation. While they displayed high visual impact, their unsuitability for medical training and scientific publications became evident. The implications of these findings are multifaceted. Primarily, the images' inaccuracies render them ineffective for training, leading to potential misconceptions. Additionally, their lack of anatomical correctness limits their applicability in scientific articles. Although the study focuses on a single AI tool, it underscores the need for collaboration between AI developers and medical professionals. The potential integration of accurate medical databases could refine the precision of such AI tools in the future. This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

Advice utilization in combined human-algorithm decision making: An analysis of preferences and behaviors

As artificial intelligence (AI) becomes more pervasive, humans will interact with autonomous agents more frequently and in deeper ways. While there is a significant body of work addressing the interface between a single human and a single AI agent, less is known about how individuals react to AI when they are part of human-agent hybrids, namely multiple humans and potentially multiple AI. These hybrid forms are unique in that advice is often given simultaneously, i.e., a human decision maker evaluates advice from other humans and algorithms at the same time. This scenario presents a boundary condition on the extant literature, as it is unclear how a human decision maker will differentially appraise a human advisor compared to an algorithmic advisor when their advice is simultaneous. This study presents the results of three experiments asking individuals to estimate property rental prices with the support of both human and algorithmic advice. We tested whether explicitly labeling an advisor as an algorithm rather than a human impacts how individuals perceive both the algorithm and another human advisor. We also examined the role of conflicting advice during simultaneous evaluation. Based on the results of 904 participants, we found that labeling an advisor as an algorithm resulted in a significantly significant algorithmic appreciation bias, even when an equivalent human was present. Further, we found that uncertainty induced by conflicting information weakened the appreciation effect, while agreement among advisors resulted in the strongest behavioral responses.

Artificial intelligence ethics code in healthcare. Sustainability of artificial intelligence systems: Why do we talk about their impact on the environment?

Environmental problems have a tremendous impact on the entire world population, particularly on human health, which plays a leading role in individual well-being. Environmental pollution, according to some estimates, kills approximately 9 million people every year. The introduction of artificial intelligence (AI) systems in many areas has enormous potential in reducing human impact on the environment; however, such systems have negative effects. The potential of AI systems to improve healthcare is inextricably linked to the ethical challenges posed by the complexity of these systems and their impact on the lives and health of communities, patients, and staff. In addition to aspects that relate directly to the algorithms, data, and clinical application of AI systems, long-term risks exist that are not obvious at first glance. One of these risks is the negative impact of AI systems on the environment, which may harm human health indirectly. AI systems are more than software, having physical components that are necessary for their functioning, such as processors, memory, and sensors. The manufacture and the energy consumption of the components has a profound effect on the environment. One study showed that when a single AI algorithm is trained, carbon emissions may reach values corresponding to the total carbon emissions from five cars lifetime.
 This study analyzes existing literature linking the development of AI systems, especially in healthcare, to their effects on the environment. The study is intended to complement the emerging AI Ethics Code for healthcare, specifically the principles of sustainability that will be included in this code.
 The study concludes that the environmental impact of AI systems should be considered when formulating ethical standards for AI in healthcare. These standards must be considered during the development, testing, and application phases of AI systems. All the people involved in the creation and use of AI systems (developers, physicians, and regulators) must monitor the environmental impact and minimize the environmental consequences of such systems at all stages of their existence. This principle calls for minimizing negative impacts, improving the energy efficiency, and disposing physical components in strict compliance with current legislation. Moreover, the rapid development of AI systems and the ethical dilemmas require that solutions be proposed jointly and ethical standards be developed in a manner that is consistent and sensitive to emerging technologies.

PO-04-151 A NEW SEQUENTIAL ARTIFICIAL INTELLIGENCE ALGORITHM FOR PREDICTING ATRIAL FIBRILLATION USING SERIAL 12-LEAD ELECTROCARDIOGRAMS

Because the early stage of atrial fibrillation (AF) is difficult to detect, a significant number of patients are exposed to the risk of stroke without forewarning. Artificial intelligence (AI) algorithms using 12-lead electrocardiograms (ECGs) are being developed to detect AF early. Unlike previous studies in which a single normal sinus rhythm (NSR) ECG was used, we hypothesized that the AI model could predict new-onset AF more accurately by learning the change between two NSR ECGs in an individual. In the present study, standard 12-lead ECGs of patients labeled by cardiologists from January 2010, to December 2021, at Samsung Medical Center were used for AI model development. All patients included in the study were classified into an NSR or AF group based on AF diagnosis. Two AI models (single ECG- and serial ECG-based) were developed using a light gradient boosting machine learning algorithm. The performance of AI models was evaluated based on the area under the receiver operating characteristic curve (AUROC) and precision-recall curve with sensitivity, specificity, accuracy, and F1 score. The AUROC results of the single ECG and serial ECG models were compared to evaluate the hypothesis. We trained 102,018 ECGs from 54,128 patients for the single ECG model and 126,106 ECGs from 106,722 patients for the serial ECG model. When testing the two AI models using test data sets, the performance for predicting new-onset AF was significantly higher with the serial AI model compared with the single AI model (single vs. serial AI model; sensitivity 0.68 vs. 0.86; specificity 0.95 vs. 0.98; accuracy 0.89 vs. 0.9; F1 score 0.74 vs. 0.90; AUROC 0.92 vs. 0.95; p<0.001). In addition, the contribution of various ECG parameters was scored using AI model developed with F-score ANOVA. In the serial ECG model, difference of P-wave amplitude between two ECGs was the leading variable for predicting AF. Comparing serial ECGs rather than a single ECG in an individual could predict future AF more accurately.Tabled 1Table 1. Baseline characteristicsSingle ECG modelOverall (n = 127,036)NSR group (n = 97,514)AF group (n = 29,522)p-valueAge, years64.3 ± 13.762.8 ± 13.570.1 ± 12.8< 0.001Male, n (%)62,136 (53.7)46,370 (50.7)15,766 (65.3)< 0.001Number of ECGs per patient1.9 ± 1.91.7 ± 1.43.3 ± 3.7< 0.001Serial ECG modelOverall (n = 157,440)NSR group (n = 129,721)AF group (n = 27,719)p-valueAge, years69.2 ± 24.367.9 ± 24.676.3 ± 20.8< 0.001Male, n (%)70,008 (50.6)56,448 (48.2)13,560 (63.4)< 0.001Number of ECGs per patient1.2 ± 1.517.5 ± 6.7< 0.001ECG=electrocardiography, NSR=normal sinus rhythm, AF=atrial fibrillation Open table in a new tab

Choose, Not Hoard: Information-to-Model Matching for Artificial Intelligence in O-RAN

Open Radio Access Network (O-RAN) is an emerging paradigm, whereby virtualized network infrastructure elements from different vendors communicate via open, standardized interfaces. A key element therein is the RAN Intelligent Controller (RIC), an Artificial Intelligence (AI)-based controller. Traditionally, all data available in the network has been used to train a single AI model to be used at the RIC. This article introduces, discusses, and evaluates the creation of multiple AI model instances at different RICs, leveraging information from some (or all) locations for their training. This brings about a flexible relationship between gNBs, the AI models used to control them, and the data such models are trained with. Experiments with real-world traces show how using multiple AI model instances that choose training data from specific locations improve the performance of traditional approaches following the hoarding strategy.

Comments: AI Language Tools Hit the Books . . . and Technical Content?

Are you among the millions of ChatGPT users/experimenters around the world? Or impatiently parked on the waitlist? OpenAI launched ChatGPT in late November 2022 with little fanfare. Within a week, the company reported 1 million users. Within the first month, it exceeded 57 million users. Since then, it has been updated several times and a UBS analyst estimated in February that the artificial intelligence (AI) language tool reached 100 million users. The website exceeded 13 million daily visitors as of January. On 1 February, OpenAI announced a ChatGPT Plus monthly subscription plan to give priority access to the chatbot. Clearly, AI language tools are fulfilling the needs and/or satisfying the curiosity of novices and more experienced AI users. Microsoft took notice and in January confirmed the extension of its partnership with OpenAI, an investment rumored to be $10 billion (not confirmed by Microsoft). Microsoft Azure will also continue as the exclusive cloud provider for the tool since OpenAI uses Azure to train its models. ChatGPT (generative pretrained transformer) is not the only AI tool available. Microsoft upgraded its Bing AI search engine, which is powered by an upgraded model of ChatGPT. ChatSonic, also built on top of ChatGPT, can access the internet. Jasper Chat, based on GPT 3.5 with OpenAI as its partner, was built for advertising and marketing businesses. Google Bard is an experimental conversational AI service powered by Google’s own next-generation language model. Character AI revolves around the concept of personas, trained with conversations in mind. Users choose from various personalities (e.g., Elon Musk, Tony Stark, Socrates, US President Biden, Kayne West, etc.) vs. interacting with a single AI chatbot. YouChat is built into a search engine and trained on a ChatGPT model. It holds conversations with full access to the internet. Caktus AI, not available as a free service, is aimed toward students and touted by the company as “the first-ever educational AI tool.” It helps with student content ranging from essays to writing paragraphs and extends to discussions, questions, and coding. There’s also a bot for programmers, GitHub Copilot X, which suggests and completes code and functions in real time. There are many more options and sure to be plenty more developed with specific user types in mind (e.g., technical specialties, business communications, conversational, special interests). All will have their own pros and cons. A retired systems engineer, formerly with General Dynamics Canada, Rob Miller, wrote a column in Canada’s National Observer about his experience using ChatGPT as a research tool for a story about carbon capture and storage—he gave a mixed review. As AI tools gain attention and provide assistance to a wide range of authors and researchers, SPE recently announced its policy for authors who use the tools to generate content for their papers. AI-generated content may be used within SPE publications, but under specific conditions. Open-source libraries, such as those used in ChatGPT and others, are not fail-safe solutions, as evidenced on 20 March when OpenAI took the chatbot offline due to a bug “which allowed some users to see titles from another active user’s chat history,” according to the company. “It’s also possible that the first message of a newly created conversation was visible in someone else’s chat history if both users were active around the same time.” OpenAI also discovered that “the same bug may have caused the unintentional visibility of payment-related information of 1.2% of the ChatGPT Plus subscribers who were active during a specific 9-hour window. … it was possible for some users to see another active user’s first and last name, email address, payment address, the last four digits (only) of a credit card number, and credit card expiration date.” I’d like to hear about your experience with ChatGPT or any of the other options. How did you use it? What type of content were you generating? If you were aiming for assistance with technical content related to oil/gas/energy, how useful was the chatbot? Share your comments and pros and cons at JPT Comments.

Approaching AI-native RANs through generalization and scalability of learning

A holistic vision of an AI-native radio-access network (RAN) would be a system designed for artificial intelligence (AI) algorithms, in which a single AI algorithm could learn and govern most networking operations, ranging from the physical layer to Radio Resource Management (RRM).

An innovative decision making method for air quality monitoring based on big data-assisted artificial intelligence technique

This work dissects the application of big data and artificial intelligence (AI) technology in environmental protection monitoring. The application principle of big data in environmental data collection is analysed based on atmospheric science and AI technology. In addition, a combined model of air quality forecasting based on machine learning is proposed to resolve real air quality monitoring challenges in environmental protection, namely, the improved complete ensemble empirical mode decomposition with adaptive noise-whale optimization algorithm-extreme learning machine (ICEEMDAN-WOA-ELM). On this basis, deep learning is introduced to establish a deep learning-based time-space-type-meteorology (TSTM) model to predict air quality. Finally, the model is verified by experiments. The results demonstrate that the ICEEMDAN-WOA-ELM model significantly outperforms a single AI model in air quality forecasting. The five evaluation index values of ICEEMDAN-WOA-ELM are 14.187, 17.235, 0.140, 0.067, and 0.946, which are higher than those of the other models. The single-step accuracy and average of the TSTM model in the heavily polluted weather forecast results almost reached full marks, with a maximum of 1.00. The performance also decreases with the growth of the step size but remains above 0.86. It can be seen that a single AI model can no longer meet the requirements of air quality forecasting. The ICEEMDAN-WOA-ELM model combined with big data has advantages in air quality monitoring and is effective for environmental protection.

Investigation of quantitative and qualitative changes in groundwater of Ardebil plain using ensemble artificial intelligence-based modeling

Abstract Groundwater is an essential source to supply water for various sectors. This paper aimed to predict the quantitative and qualitative changes in groundwater over time and to evaluate the efficiency of different modeling methods. This study is based on three steps. In the first step, quantitative and qualitative piezometers were clustered by the Growing Neural Gas Network (GNG) method, and the central piezometer of each cluster was used on behalf of each cluster. In the second step, four different Artificial Intelligence (AI) models were applied, namely Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), Support Vector Machine (SVM), and Emotional Artificial Neural Network (EANN). As a post-processing approach three different ensemble methods were used: simple average ensemble (SAE), weighted average ensemble (WAE), and nonlinear neural network ensemble (NNE). In the third step, the outputs of single AI models were used to enhance the evaluation results. Therefore, the results demonstrate that the NNE led to reach the better performance for three GWL, TDS, and TH parameters up to 37, 29, and 23% on average, respectively. Study results will lead to the improvement of AI applications in groundwater research and will benefit groundwater development plans.

Evaluation of the Combination of Artificial Intelligence and Radiologist Assessments to Interpret Malignant Architectural Distortion on Mammography.

PurposeTo compare the mammographic malignant architectural distortion (AD) detection performance of radiologists who read mammographic examinations unaided versus those who read these examinations with the support of artificial intelligence (AI) systems.Material and MethodsThis retrospective case-control study was based on a double-reading of clinical mammograms between January 2011 and December 2016 at a large tertiary academic medical center. The study included 177 malignant and 90 benign architectural distortion (AD) patients. The model was built based on the ResNeXt-50 network. Algorithms used deep learning convolutional neural networks, feature classifiers, image analysis algorithms to depict AD and output a score that translated to malignant. The accuracy for malignant AD detection was evaluated using area under the curve (AUC).ResultsThe overall AUC was 0.733 (95% CI, 0.673-0.792) for Reader First-1, 0.652 (95% CI, 0.586-0.717) for Reader First-2, and 0.655 (95% CI, 0.590-0.719) for Reader First-3. and the overall AUCs for Reader Second-1, 2, 3 were 0.875 (95% CI, 0.830-0.919), 0.882 (95% CI, 0.839-0.926), 0.884 (95% CI, 0.841-0.927),respectively. The AUCs for all the reader-second radiologists were significantly higher than those for all the reader-first radiologists (Reader First-1 vs. Reader Second-1, P= 0.004). The overall AUC was 0.792 (95% CI, 0.660-0.925) for AI algorithms. The combination assessment of AI algorithms and Reader First-1 achieved an AUC of 0.880 (95% CI, 0.793-0.968), increased than the Reader First-1 alone and AI algorithms alone. AI algorithms alone achieved a specificity of 61.1% and a sensitivity of 80.6%. The specificity for Reader First-1 was 55.5%, and the sensitivity was 86.1%. The results of the combined assessment of AI and Reader First-1 showed a specificity of 72.7% and sensitivity of 91.7%. The performance showed significant improvements compared with AI alone (p<0.001) as well as the reader first-1 alone (p=0.006).ConclusionWhile the single AI algorithm did not outperform radiologists, an ensemble of AI algorithms combined with junior radiologist assessments were found to improve the overall accuracy. This study underscores the potential of using machine learning methods to enhance mammography interpretation, especially in remote areas and primary hospitals.

Interactive Design Psychology and Artificial Intelligence-Based Innovative Exploration of Anglo-American Traumatic Narrative Literature.

The advent of the intelligence age has injected new elements into the development of literature. The synergic modification of Anglo-American (AAL) traumatic narrative (TN) literature by artificial intelligence (AI) technology and interactive design (ID) psychology will produce new possibilities in literary creation. First, by studying natural language processing (NLP) technology, this study proposes a modification language model (LM) based on the double-layered recurrent neural network (RNN) algorithm and constructs an intelligent language modification system based on the improved LM model. The results show that the performance of the proposed model is excellent; only about 30% of the respondents like AAL literature; the lack of common cultural background, appreciation difficulties, and language barriers have become the main reasons for the decline of reading willingness of AAL literature. Finally, AI technology and ID psychology are used to modify a famous TN work respectively and synergically, and the modified work is appreciated by respondents to collect their comments. The results corroborate that 62% of the respondents like original articles, but their likability scores have decreased for individually modified work by AI or ID psychology. In comparison, under the synergic modification efforts of AI and ID psychology, the popularity of the modified work has increased slightly, with 65% of the respondents showing a likability to read. Therefore, it is concluded that literary modification by single ID psychology or AI technology will reduce the reading threshold by trading off the literary value of the original work. The core of literary creation depends on human intelligence, and AI might still not be able to generate high-standard literary works independently because human minds and thoughts cannot be controlled and predicted by machines. The research results provide new ideas and improvement directions for the field of AI-assisted writing.

Doctor Ex Machina: A Critical Assessment of the Use of Artificial Intelligence in Health Care.

This article examines the potential implications of the implementation of artificial intelligence (AI) in health care for both its delivery and the medical profession. To this end, the first section explores the basic features of AI and the yet theoretical concept of autonomous AI followed by an overview of current and developing AI applications. Against this background, the second section discusses the transforming roles of physicians and changes in the patient-physician relationship that could be a consequence of gradual expansion of AI in health care. Subsequently, an examination of the responsibilities physicians should assume in this process is explored. The third section describes conceivable practical and ethical challenges that implementation of a single all-encompassing AI healthcare system would pose. The fourth section presents arguments for regulation of AI in health care to ensure that these applications do not violate basic ethical principles and that human control of AI will be preserved in the future. In the final section, fundamental components of a moral framework from which such regulation may be derived are brought forward, and some possible strategies for building a moral framework are discussed.

Toward Generalizability in the Deployment of Artificial Intelligence in Radiology: Role of Computation Stress Testing to Overcome Underspecification.

The clinical deployment of artificial intelligence (AI) applications in medical imaging is perhaps the greatest challenge facing radiology in the next decade. One of the main obstacles to the incorporation of automated AI-based decision-making tools in medicine is the failure of models to generalize when deployed across institutions with heterogeneous populations and imaging protocols. The most well-understood pitfall in developing these AI models is overfitting, which has, in part, been overcome by optimizing training protocols. However, overfitting is not the only obstacle to the success and generalizability of AI. Underspecification is also a serious impediment that requires conceptual understanding and correction. It is well known that a single AI pipeline, with prescribed training and testing sets, can produce several models with various levels of generalizability. Underspecification defines the inability of the pipeline to identify whether these models have embedded the structure of the underlying system by using a test set independent of, but distributed identically, to the training set. An underspecified pipeline is unable to assess the degree to which the models will be generalizable. Stress testing is a known tool in AI that can limit underspecification and, importantly, assure broad generalizability of AI models. However, the application of stress tests is new in radiologic applications. This report describes the concept of underspecification from a radiologist perspective, discusses stress testing as a specific strategy to overcome underspecification, and explains how stress tests could be designed in radiology-by modifying medical images or stratifying testing datasets. In the upcoming years, stress tests should become in radiology the standard that crash tests have become in the automotive industry. Keywords: Computer Applications-General, Informatics, Computer-aided Diagnosis © RSNA, 2021.

Systematic review of application of artificial intelligence tools in architectural, engineering and construction

PurposeThis study reviews the extent of application of artificial intelligence (AI) tools in the construction industry.Design/methodology/approachA thorough literature review (based on 165 articles) was conducted using Elsevier's Scopus due to its simplicity and as it encapsulates an extensive variety of databases to identify the literature related to the scope of the present study.FindingsThe following items were extracted: type of AI tools used, the major purpose of application, the geographical location where the study was conducted and the distribution of studies in terms of the journals they are published by. Based on the review results, the disciplines the AI tools have been used for were classified into eight major areas, such as geotechnical engineering, project management, energy, hydrology, environment and transportation, while construction materials and structural engineering. ANN has been a widely used tool, while the researchers have also used other AI tools, which shows efforts of exploring other tools for better modelling abilities. There is also clear evidence of that studies are now growing from applying a single AI tool to applying hybrid ones to create a comparison and showcase which tool provides a better result in an apple-to-apple scenario.Practical implicationsThe findings can be used, not only by the researchers interested in the application of AI tools in construction, but also by the industry practitioners, who are keen to further understand and explore the applications of AI tools in the field.Originality/valueThere are no studies to date which serves as the center point to learn about the different AI tools available and their level of application in different fields of AEC. The study sheds light on various studies, which have used AI in hybrid/evolutionary systems to develop effective and accurate predictive models, to offer researchers and model developers more tools to choose from.

Multi-station artificial intelligence based ensemble modeling of suspended sediment load

Abstract In this study, Artificial Intelligence (AI) models along with ensemble techniques were employed for predicting suspended sediment load (SSL) via single station and multi-station scenarios. Feed Forward Neural Networks (FFNNs), Adaptive Neuro-Fuzzy Inference System (ANFIS), and Support Vector Regression (SVR) were the employed AI models, and simple averaging (SA), weighted averaging (WA), and neural averaging (NA) were the ensemble techniques developed for combining the outputs of the individual AI models to gain more accurate estimations of the SSL. For this purpose, twenty-year observed streamflow and SSL data of three gauging stations, located in the Missouri and Upper Mississippi regions were utilized at both daily and monthly scales. The obtained results of both scenarios indicated the supremacy of ensemble techniques to single AI models. The neural ensemble demonstrated more reliable performance compared to other ensemble techniques. For instance, in the first scenario, the ensemble technique increased the predicted results up to 20% in the verification phase of the daily and monthly modeling and up to 5 and 8%, respectively, in the verification step of the second scenario.

Applications of artificial intelligence in engineering and manufacturing: a systematic review

Engineering and manufacturing processes and systems designs involve many challenges, such as dynamism, chaotic behaviours, and complexity. Of late, the arrival of big data, high computational speed, cloud computing and artificial intelligence techniques (like machine learning and deep learning) has reformed how many engineering and manufacturing professionals approach their work. These technologies offer thrilling innovative ways for engineers and manufacturers to tackle real-life challenges. On the other hand, the field of Artificial Intelligence (AI) is extensive. Several diverse theories, algorithms, and methods are available, which presents a challenge and a barrier in choosing the right AI technique for the appropriate engineering process or manufacturing process and environments. Besides, the pertinent literature is disseminated over various journals, conference proceedings, and research communities. Hence, conducting a systematic survey to scrutinise and classify the existing literature is worthwhile. However, it is challenging, but previous review studies have not adequately addressed AI’s use and advancement in engineering and manufacturing (EM). Besides, some concentrated on single AI models, and others focused on a specific area in EM. This paper presents a comprehensive systematic review of studies on AI and its application in EM. To limit the scope of the current study, we conducted a keyword search in official publisher websites and academic databases, such as Springer, Elsevier, Scopus, Science Publication, Taylor & Francis, Directory of Open Access Journals (DOAJ), Association for Computing Machinery (ACM), Wiley online library, Inderscience and Google scholar. The search results (173 articles) were filtered according to a proposed framework, which resulted in ninety-one (91) relevant research articles. We reviewed the articles based on a proposed taxonomy (the year of publication, the AI algorithm and machine learning task adopted, the application area in EM, the train and test split of data, the error, and accuracy metrics used, the potential benefits). Our assessment using the proposed taxonomy gave a helpful insight into the literature’s anatomy on various AI applications in engineering and manufacturing. Also, we identified opportunities for future research in AI application in the field of EM.

Metaheuristics‐optimized ensemble system for predicting mechanical strength of reinforced concrete materials

This paper develops a novel artificial intelligence (AI)-based approach, called the metaheuristics-optimized ensemble system (MOES), to assist civil engineers significantly in achieving accurate estimations of the mechanical strength of reinforced concrete (RC) materials. MOES integrates the advantages of hybrid and ensemble models by combining a metaheuristic optimization algorithm and efficient AI models. The metaheuristic algorithm finds the optimal hyperparameters of individual AI techniques and simultaneously adjusts their weights to yield the best optimized-weight-ensemble model. Particularly, the developed MOES was established by integrating the forensic-based investigation optimization algorithm, the radial basis function neural network, and the least squares support vector regression. Four case studies of predicting structural mechanics of RC beams were performed to evaluate the performance of MOES and compare it to those of other single AI models, conventional ensemble models, hybrid models, and empirical methods. The analytical results of cross-validation reveal that MOES was the most reliable approach, achieving the best values of all performance evaluation indexes. The automated predictive analytics revealed the robustness, efficiency, and stability of MOES. Thus, the proposed approach is a highly promising tool for predicting the structural mechanics of RC beams. The success of MOES in estimating the mechanical strength of RC beams has redefined the way of optimizing an ensemble AI model, which is the primary contribution of this research to the relevant body of knowledge.