Federated Learning Lifecycle Management for Distributed Medical Artificial Intelligence Applications: A Case Study on Post-Transcatheter Aortic Valve Replacement Complication Prediction Solution

Aim: The majority of machine learning (ML) models in healthcare are built on retrospective data, much of which is collected without explicit patient consent for use in artificial intelligence (AI) and ML applications. The primary aim of this study was to evaluate whether clinicians and scientific researchers themselves consent to provide their own data for the training of ML models. Materials and Methods: This survey was conducted through an anonymous online survey, utilizing platforms such as Telegram, LinkedIn, and Viber. The target audience comprised specific international groups, primarily Russian, German, and English-speaking, of clinicians and scientific researchers. These participants ranged in their levels of expertise and experience, from beginners to veterans. The survey centered on a singular, pivotal question: “Do You Consent to the Use of Your Biological and Private Data for Training Machine Learning and AI Models?” Respondents had the option to choose from three responses: “Yes” and “No”. Results: The survey was conducted in January 2024. A total of 119 unique and verified individuals participated in the survey. The results revealed that only 50% of respondents (63 persons) expressed consent to provide their own data for the training of ML and AI models. Conclusion: In the development of ML and AI models, particularly open-source ones, it is crucial to ascertain whether participants are willing to provide their private data. While ML algorithms can transform the nature of data, it is important to remember that the primary owner of this data is the individual. Our findings show that in 50% of the cases, even participants from scientific research and clinical backgrounds – individuals typically accountable for ensuring data quality in AI and ML model development – do not consent to the use of their data in AI and ML settings. This highlights the need for more stringent consent processes and ethical considerations in the utilization of personal data in AI and ML research.

Over the last decade, we've seen a quick uptake of artificial intelligence (AI) solutions in publishing.AI is used to facilitate a range of tasks, from presubmission text revision to post-editing quality control.But while most AI solutions are generic, most publishers have their own norms when it comes to language, style, and editing.Can an AI solution account for such publisherspecific needs? AI concerns at the ISMTE Global EventThe last ISMTE Global Event in November 2022 included a session titled "I AM EDITOR -AI in publishing."In it, I presented Writefull's AI services for publishers, like its APIs for copy editing and language scoring.After the talk, some attendees wondered about the limitations of AI, in particular considering its generic output.Some of the questions asked were:Attendees' main concern seemed to be: to what extent can an AI language model meet publisher-specific needs, such as following style guides or refraining from changing the author's voice?This reflects the position of many publishers with whom we work."Yes, we would like to use an AI solution to facilitate our work, but we only want it to do A and B-and not C and D." This is a valid concern.In the end, if an AI cannot follow instructions, if it doesn't align very closely with what a human copy editor or scorer would do, it may hinder rather than facilitate.And this is the crux.An AI model can be trained on texts before and after copy editing, but unless the training data are tailored to the needs of a particular publisher, its output will always be generic.It will learn to fix common language errors, but the edits may be too few, too many, or simply stylistically different from what any given publisher appreciates.

Healthcare 5.0 shows an evolution in intelligent, integrated, and patient-centric health systems, which is based on the interconnected use of technologies such as Internet of Medical Things (IoMT), Artificial Intelligence (AI) and edge computing. But, with increasing dependence on these technologies, it also causes several concers, such as data privacy, security, and interoperability. The sharing of data at one centre for training can increase the risk of data breach, raising worries about data safety. Federated learning, on the other hand, allows training of machine learning models without sharing data in a decentralised manner. The patient privacy and data remain protected as sensitive data can remain on the device. Simultaneously, Blockchain improves this framework by providing a secure and transparent record of data transactions, providing accurate data tracking which further reduces the risk of data tampering. In this paper, we will evaluate the integration of Federated learning with blockchain technology in the framework of Internet of Medical Things (IoMT). We present a system design which guarantees secure data exchange, reliable model aggregation and traceability. Additionally, We will also discuss about challenges and ethical consideration in its implementation.

Despite tremendous efforts, including research, teaching, and extension, toward improving the safety of agricultural tractor drivers, the number of incidents related to agricultural tractor drivers has not declined. This evidence points out an urgent need to explore artificial intelligence (AI) solutions to improve the safety of tractor drivers. This paper uses 171 Fatality Assessment and Control Evaluation (FACE) reports related to tractor drivers and a new framework called Risk Evolution, Detection, Evaluation, and Control of Accidents (REDECA) to identify existing AI solutions, such as machine learning models for predictive maintenance, sensor-based monitoring, computer vision, and automated safety interventions, and specific areas where AI solutions are missed and can be developed to reduce incidents and recovery time. Fatality reports of tractor drivers were categorized into six main categories, including run over, pinned by/ Crushed and entanglement, fall, fire, roll over, and overturn. Each category was then subcategorized based on similarities of incident causes in the reports. The application of the REDECA framework, which categorizes risk states into R1 (safe), R2 (hazard exposure), and R3 (incident), revealed potential AI solutions that could improve the safety of tractor drivers. In all categories, the REDECA framework lacks AI solutions for three elements, including the probability of reducing recovery time in R3, detecting changes between R2 and R3, and intervention to send workers to R2. Most of the categories were missing AI solutions for interventions to prevent entry to the R3 element of the REDECA. In addition, the fall, roll over, and overturn categories lacked AI intervention that minimized damage and recovery in R3. The outcome of this study shows an urgent need to develop AI solutions to improve tractor driver safety.

Added value of artificial intelligence solutions for arterial stenosis detection on head and neck CT angiography: A randomized crossover multi-reader multi-case study

Artificial intelligence (AI) has the potential to revolutionise public service delivery, but its creation and implementation come with both exciting possibilities and complex obstacles. Despite the increasing discussions surrounding AI, there is a lack of research specifically focused on its implications within public service sectors, healthcare, education, and transportation. This paper aims to fill this gap by critically examining how AI can enhance efficiency, decision-making, and service accessibility while also exploring the hurdles posed by its implementation, including technical infrastructure requirements, workforce adaptation, ethical concerns, and governance complexities. By conducting a thorough analysis of existing research, the study uncovers significant opportunities, including enhanced accessibility, data-driven insights, and streamlined operations. At the same time, it emphasises significant challenges, such as algorithmic bias, data privacy risks, public trust deficits, and resource disparities that may impede the equitable adoption of AI. To tackle these obstacles and promote responsible AI deployment, this paper examines strategic approaches that encompass establishing transparent governance frameworks to ensure accountability, enhancing data privacy and security protocols to safeguard public information, and fostering AI literacy through comprehensive workforce training. Moreover, it seeks to echo the importance of ethical AI development aimed at addressing bias and fostering inclusivity, while ensuring that AI solutions are in line with societal requirements. Building scalable and inclusive AI infrastructure is crucial for closing the digital divide and guaranteeing fair access to AI-powered services, especially in marginalised communities. Additionally, building public trust by implementing transparent policies and actively involving citizens in the adoption process is vital for the successful integration of AI. Given the scarcity of research on AI’s impact on public service delivery, this paper offers valuable insights to inform policymakers, public administrators, and stakeholders in effectively navigating the challenges associated with AI adoption. By providing a detailed analysis of the advantages and potential challenges, and by suggesting effective implementation strategies, the study seeks to improve governance practices in the digital era, promoting fairness and inclusivity.

The advent of artificial intelligence (AI) presents a transformative opportunity to revolutionize agriculture and rural economies, offering a pathway to sustainable livelihoods. In precision agriculture, AI enables real-time monitoring of soil conditions, crop health, and weather patterns, and optimizes resource use through AI-driven irrigation and fertilization systems. AI-based image recognition and predictive models allow farmers for early detection of pest and disease outbreaks, leading to targeted interventions that reduce environmental impact and costs. In addition, supply chain management and market access benefit from AI algorithms that forecast market trends, optimize logistics and reduce post-harvest losses. Community empowerment is achieved by ensuring inclusive participation in AI solutions and creating job opportunities in AI and technology-related roles. To fully harness the potential of AI, supportive policies and robust infrastructure are essential, along with addressing ethical considerations such as data privacy and equitable access.

This paper analyzes the optimization features of machine learning (ML) model training procedures using multi-GPU systems to enhance cyber security in telecommunication networks. A key aspect of the study is the use of data parallelism, which allows the distribution of the training load across multiple GPUs, significantly reducing training time and improving model accuracy-critical factors for rapid threat detection in cyberspace. A novel approach for optimizing data batch size using Mutual Information (MI) is proposed, which harmonizes the utilization of computational resources with the information content of the training data. MI helps to determine the optimal data batch size that minimizes training errors and improves model accuracy without a significant increase in training time. Experimental results demonstrate the substantial advantages of multi-GPU configurations compared to single-GPU setups, providing faster training and improved model accuracy. It was particularly emphasized that MI-guided batch size tuning significantly outperforms traditional manual tuning methods, ensuring higher validation accuracy and reducing training time. The study showed that the MI-based approach is an effective tool for addressing the problem of optimizing ML model training processes in real-world scenarios where cyber security is critical. The proposed methods allow ML models to train faster and more accurately identify potential threats, making them particularly relevant for telecommunication networks where a rapid response to new threats in real time is required. The implementation of modern computational technologies such as multi-GPU systems and MI-optimized training enhances the efficiency and accuracy of machine learning models. This, in turn, improves cyber security measures and ensures a more reliable defence of telecommunication networks against malicious attacks. It is noted that the proposed approaches can be adapted not only for cyber security but also for other areas where high model accuracy and fast training are important. Future research prospects include the development of new machine learning methods, particularly deep neural networks, the exploration of alternative computational architectures such as quantum computing or distributed systems, and their integration into real-time systems. Special attention should be paid to the ethical aspects of implementing automated cyber security systems, particularly in preventing bias in algorithms and ensuring fairness in their application.

Data privacy is a very important issue. Especially in fields like medicine, it is paramount to abide by the existing privacy regulations to preserve patients’ anonymity. However, data is required for research and training machine learning models that could help gain insight into complex correlations or personalised treatments that may otherwise stay undiscovered. Those models generally scale with the amount of data available, but the current situation often prohibits building large databases across sites. So it would be beneficial to be able to combine similar or related data from different sites all over the world while still preserving data privacy. Federated learning has been proposed as a solution for this, because it relies on the sharing of machine learning models, instead of the raw data itself. That means private data never leaves the site or device it was collected on. Federated learning is an emerging research area, and many domains have been identified for the application of those methods. This systematic literature review provides an extensive look at the concept of and research into federated learning and its applicability for confidential healthcare datasets.

The future 6G network will be able to connect massive and various unmanned vehicles (UxVs) into the network, bringing novel requirements toward UxV network security and data privacy. Deploying machine learning (ML)-based intrusion detection on UxVs can be one promising approach. The conventional approach cannot fulfill security and privacy requirements, as the structure can be more flexible, and there may be a lack of a central node in the 6G-based UxV network. Moreover, UxV clients can exhibit high heterogeneity with extremely class-imbalanced training samples. With the presence of federated learning (FL), UxVs can collaborate to train and update ML models while preserving data privacy in the UxV security critical scenario. However, most past works focused on centralized approaches, using cloud servers to assist intrusion detection and service deployments. This article introduces an FL framework with a decentralized design for training ML models on UxVs using intrusion detection as a study case. This scheme can be more flexible, as UxV clients can collaborate to train and synchronize models without the central server. Simulation experiments were conducted to evaluate the efficiency of the proposed approach. Experimental results and theoretical analysis confirm that the proposed method outperforms baseline models with FedAvg and local-trained client models.

Artificial intelligence (AI) has huge potential to accelerate drug discovery, but challenges remain in implementing AI algorithms that can be used by the broad scientific community. Identification of molecular features and their subsequent use in training of machine learning models may permit prediction of new molecules with enhanced properties. Predictive modeling is particularly applicable to analysis of structure-activity relationships (SARs) and would be a useful tool in the hands of laboratory medicinal chemists. This requires a software platform that is chemically intuitive while providing the user with access to AI methods. The KNIME platform provides such an environment through inclusion of broad chemical toolsets and a user-friendly approach for utilization of machine learning for analysis of SAR data. Here, we illustrate use of KNIME for this purpose, with a focus on discovery of features of highly potent tau inhibitors from a series of structurally diverse polyphenols. Workflows are described that enable implementation of AI tools in KNIME for diverse SAR projects.

Artificial Intelligence (AI) has a significant impact on empowering autonomous driving systems to perceive and interpret the environment effectively. However, ensuring data privacy and security in autonomous driving systems is a critical challenge. To surmount these hurdles, federated learning has emerged as an effective strategy. Federated learning is a decentralized machine learning approach that facilitates the cooperative training of models across a diverse set of connected devices, enabling them to collectively learn and improve their performance, while preserving data privacy. This approach eliminates the necessity of sharing raw data and only involves sharing model updates with a central aggregator, thereby ensuring privacy and minimizing data exposure. This paper examines the implementation of federated learning in autonomous driving. It explores the principles of federated learning, including decentralized training, local model updates, model aggregation, privacy preservation, iterative learning, and heterogeneity handling. Two specific approaches, Deep Federated Learning (DFL) and End-to-End Federated Learning, are discussed, highlighting their benefits in enhancing privacy and maintaining prediction accuracy. The paper also discusses the applications of federated learning in communication and control aspects of autonomous driving. It emphasizes the scalability, adaptability, edge computing, real-time learning, federated transfer learning, and privacy-preserving data sharing as potential future prospects for federated learning in autonomous driving. Overall, federated learning offers a unique opportunity to address privacy concerns in autonomous driving systems while harnessing the collective intelligence of a fleet of vehicles. It has the potential to revolutionize the field and contribute to the development of safe and secure autonomous driving technologies.

Integrating artificial intelligence (AI) into health care offers the potential to address critical challenges related to access to care, workforce burnout, and health inequities. Despite its promise, AI adoption remains limited due to safety, efficacy, and equity concerns. This paper presents a novel and comprehensive framework for responsible AI development, evaluation, and deployment in health care, encompassing four key phases: (1) AI Solution Design and Development, (2) AI Solution Qualification, (3) AI Solution Efficacy and Safety Evaluation, and (4) AI Solution Impact. By establishing rigorous standards for operational, clinical, and technical quality, the framework aims to guide AI developers and health care professionals toward creating AI solutions that are ethical, effective, and scalable. This structured approach fosters collaboration and mitigates risks to help AI achieve its full potential in improving patient outcomes and health care efficiency.

This article aims to analyse Articles 5 and 6 of the draft ePrivacy Regulation put forward by the European Commission, as key rules regulating the processing of electronic communication data and metadata. The confidentiality of electronic communication is an important aspect of privacy and personal autonomy protection. Still, disproportionate regulation may hurt economic growth, particularly with regard to artificial intelligence (AI) solutions development. The article begins by briefly describing a socio-economic context in which the future regulation of electronic communication confidentiality will function, then analyses the implications of proposed norms for the protection of privacy and personal autonomy, and their potential implications for economic development, for AI solutions in particular. The article analyses which of the proposed versions of Articles 5 and 6 meet the middle ground and ensure protection of privacy and personal autonomy without at the same time hampering economic development and AI innovation. After analysing the proposed normative content of all three versions of the ePrivacy Regulation draft, some afterthoughts are shared about them and their potential impact. The goal is to find the proper balance between privacy protection as an ultimate priority and maintaining economic development and innovation as something that cannot be ignored and is a priority in its own right, to an extent where it does not harm the essential content of the fundamental right to privacy and personal autonomy.

Purpose This study explores the adoption of artificial intelligence (AI) in wineries, with a specific focus on its application to drive innovation and on the organisational and contextual factors that influence its exploration and adoption. Design/methodology/approach The research project is based on an exploratory approach employing a questionnaire developed through a literature review and refined using the Delphi method via a survey with over 500 participants. Findings Wineries employ many AI solutions, from generative AI tools that facilitate creative and agile processes to more embedded, enterprise-level AI systems that require significant investment and IT integration. They adopt a dual approach to exhibit the highest innovation orientation by integrating AI solutions to promote many dimensions of innovation. In contrast, wineries that rely exclusively on generative AI leverage it to innovate marketing processes. However, most wineries surveyed have not implemented AI solutions for process innovation, suggesting that AI development and adoption are in their infancy. Research limitations/implications The study did not examine whether the adoption of AI solutions actually generated innovation, nor was the extent of AI adoption at the micro or meso level assessed. However, the investigations may provide valuable insights and contribute to the development of more targeted strategies for the digital transformation of the wine sector. By exploring organisational and contextual dimensions, it provides a deeper understanding of the variables that may encourage or hinder the implementation of AI technologies. Originality/value This study can be leveraged by wineries as a pioneering analysis of innovation-oriented AI adoption in the wine sector in several countries. The findings offer practical value for the wine industry in promoting digital transformation and harnessing the innovative potential of AI in the wine sector.