Machine learning (ML) in medicine has transitioned from research to concrete applications aimed at supporting several medical purposes like therapy selection, monitoring and treatment. Acceptance and effective adoption by clinicians and patients, as well as regulatory approval, require evidence of trustworthiness. A major factor for the development of trustworthy AI is the quantification of data quality for AI model training and testing. We have recently proposed the METRIC-framework for systematically evaluating the suitability (fit-for-purpose) of data for medical ML for a given task. Here, we operationalize this theoretical framework by introducing a collection of data quality metrics - the metric library - for practically measuring data quality dimensions. For each metric, we provide a metric card with the most important information, including definition, applicability, examples, pitfalls and recommendations, to support the understanding and implementation of these metrics. Furthermore, we discuss strategies and provide decision trees for choosing an appropriate set of data quality metrics from the metric library given specific use cases. We demonstrate the impact of our approach exemplarily on the PTB-XL ECG-dataset. This is a first step to enable fit-for-purpose evaluation of training and test data in practice as the base for establishing trustworthy AI in medicine.
This study examines how physicians and nurses in Brazil are using generative artificial intelligence tools in healthcare practice. Based on data from the ICT in Health 2024 survey, findings reveal that AI adoption remains limited and uneven across professional categories and healthcare settings. Nurses use AI mainly for research and communication, while physicians apply it to clinical documentation and research. The results highlight institutional and infrastructural disparities that influence adoption and point to the need for training, ethical governance, and stronger digital capacity to foster equitable and human-centred AI integration in healthcare.
The dAIEDGE Network of Excellence (NoE) seeks to strengthen and support the development of a dynamic European cutting-edge Artificial intelligence (AI) ecosystem under the umbrella of the European Lighthouse for AI, and to sustain the development of advanced AI. dAIEDGE fosters the exchange of ideas, concepts, and trends on cutting-edge next generation AI, creating links between ecosystem actors to help both the European Commission (EC) and the European Union (EU) and the peripheral AI constituency identify strategies for future developments in Europe. Our main objective is to advance Europe’s innovation and technology base by developing a comprehensive policy and governance approach to AI in order for the EU to become a world leader in innovation in the data economy and its applications.
Crystalline silicon carbon nitrite thin films have been synthesized by continuous MPACVD in N2/CH4 gas mixture on silicon substrates. Prior deposition, discharges stabilisations have been studied and then the plasma is analysed in-situ by optical emission spectroscopy according to experimental parameters (microwave power, gas mixture, pressure and flow rate). Several techniques have been used to characterize the films. Morphological analyses realized by SEM, TEM and AFM show that the films are nano-crystalline and their roughness is around 5 nm. The XRD spectrum and selected area electron diffraction patterns exhibit a signature that correspond to beta-C3N4 and SiCN. EDXS and XPS analyses show the presence of C, N, Si and O. The C1s, N1s and Si2p levels obtained from XPS confirm the presence of C-N, Si-C and Si-N covalent bonds typical for SiCN and CNx films. The presence of Si is justified by the participation of the substrate to the growth by etching. The oxygen comes from the exposure of the films to atmosphere after deposition. The surface was not cleaned by argon bombardment to avoid a structural and bonding modification of the film.
The rapid evolution of artificial intelligence (AI), particularly in convolutional neural networks (CNNs) and deep learning, has revolutionized numerous domains, ranging from medical imaging to creative arts and legal analytics. This research emphasizes the role of pre-trained CNN architectures in identifying kidney conditions, leveraging a dataset comprising images of healthy kidneys as well as those affected by cysts, tumors, and stones. The pretrained models known for their outstanding image recognition capabilities, were adapted for this classification task through transfer learning (TL) techniques. By refining these models and carefully calibrating key parameters like learning rate, batch size, and network depth, they demonstrated superior performance compared to traditional machine learning approaches. The findings underscore the transformative potential of pre-trained CNNs in advancing the precision of kidney disease diagnostics, with implications for broader medical applications.
