Objective:Medication errors remain a substantial public health concern, and existing measures, such as workforce training, have achieved only partial success. Advances in data availability and computational methods have led to increasing use of machine learning (ML) to support medication safety. This scoping review synthesizes and categorizes ML-based approaches to medication error detection or prediction. Materials and Methods:Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines, PubMed, Embase, and Web of Science were searched for studies published between 2015 and April 2025. Two reviewers independently performed study selection using predefined eligibility criteria, and data extraction followed a structured extraction framework. Results:Twenty-two studies met the inclusion criteria. Two dominant ML pipelines were identified. Most studies focused on prescription-related errors, relying on structured clinical data and tree-based models. A smaller group addressed medication-administration errors using unstructured multimodal data, such as images or video, analyzed with neural networks and multi-stage detection pipelines. Discussion:ML shows substantial potential for medication error detection, particularly in prescription-focused workflows that align well with existing clinical processes. However, the evidence remains fragmented, with limited generalizability, inconsistent labeling, and scarce real-world evaluation. No studies addressed medication errors in clinical research settings, such as clinical trials, despite their distinct workflows and safety implications. Conclusion:Advancing ML-based medication error detection will require high-quality multicenter datasets, rigorous and transparent validation, and deeper exploration of underused data modalities, including free text.

Radiological reports are essential for clinical diagnosis. However, their preparation is laborious and time-consuming for radiologists. Artificial intelligence (AI) has the potential to help radiologists reduce this workload. This scoping review aims to map current research on AI-based radiological report generation, highlighting its applications, limitations, and future directions. We systematically retrieved 321 records from three major scientific databases (PubMed, Embase, Web of Science) and included 58 studies after screening. Benchmark datasets, radiological subspecialties, AI models, output formats, and evaluation methods were extracted and analyzed. Nine public benchmark datasets and six radiological subspecialties were identified. Transformer-based architectures have become the dominant approach, with outputs primarily generated in natural language. Dataset diversity is limited, with most studies relying on chest X-ray images from two public datasets. Structured outputs for machine readability remain underexplored, and expert human evaluation is infrequently used as an assessment method. Current AI research in radiological report generation is constrained by limited dataset diversity and evaluation practices. Future studies should expand imaging modalities and subspecialties, adopt unified vision-language Transformer models, generate more structured outputs for programmatic use, and incorporate expert human evaluation for quality assessment.

Objective: The objective of this study is to develop a machine learning (ML)-based framework for early risk stratification of clinical trials (CTs) according to their likelihood of exhibiting a high rate of dosing errors, using information available prior to trial initiation. Materials and Methods: We constructed a dataset from ClinicalTrials.gov comprising 42,112 CTs. Structured, semi-structured trial data, and unstructured protocol-related free-text data were extracted. CTs were assigned binary labels indicating elevated dosing error rate, derived from adverse event reports, MedDRA terminology, and Wilson confidence intervals. We evaluated an XGBoost model trained on structured features, a ClinicalModernBERT model using textual data, and a simple late-fusion model combining both modalities. Post-hoc probability calibration was applied to enable interpretable, trial-level risk stratification. Results: The late-fusion model achieved the highest AUC-ROC (0.862). Beyond discrimination, calibrated outputs enabled robust stratification of CTs into predefined risk categories. The proportion of trials labeled as having an excessively high dosing error rate increased monotonically across higher predicted risk groups and aligned with the corresponding predicted probability ranges. Discussion: These findings indicate that dosing error risk can be anticipated at the trial level using pre-initiation information. Probability calibration was essential for translating model outputs into reliable and interpretable risk categories, while simple multimodal integration yielded performance gains without requiring complex architectures. Conclusion: This study introduces a reproducible and scalable ML framework for early, trial-level risk stratification of CTs at risk of high dosing error rates, supporting proactive, risk-based quality management in clinical research.

Introduction Medication errors pose a significant threat to public health. Despite efforts by health agencies and the implementation of various interventions, such as staff training, medication reconciliation and automation, the persistence of these incidents highlights the need for more effective, scalable solutions. In recent years, machine learning (ML) has emerged as a promising approach in healthcare, offering potential to detect and predict medication errors through data-driven insights. This scoping review aims to systematically map the existing literature on ML-based approaches to predict or detect medication errors across all stages of the medication use process. The review seeks to identify the range of ML applications in this domain, characterise methodological trends and highlight current knowledge gaps. The findings will provide a structured and accessible overview for both clinicians and researchers, supporting the development of safer, more data-informed medication practices.Methods and analysis The review will be conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guideline. Structured searches will be performed in PubMed, Embase and Web of Science, covering publications from 1 January 2015 to 28 April 2025. Predefined inclusion and exclusion criteria will be used to identify eligible studies. Key information—including ML models, data sources and type, evaluation methods and clinical contexts—will be extracted and analysed using descriptive statistics, visualisations, thematic analysis and narrative synthesis.Ethics and dissemination This study involves a review of existing literature and does not involve human participants, personal data or unpublished secondary data. As such, ethical approval was not required. All data analysed were obtained from publicly available sources. Findings of the scoping review will be disseminated through professional networks, conference presentations and publications in scientific journals.Trial registration number This protocol has been registered on the Open Science Framework (https://doi.org/10.17605/OSF.IO/38SFY).

OBJECTIVE:The primary objective of our study is to address the challenge of confidentially sharing medical images across different centers. This is often a critical necessity in both clinical and research environments, yet restrictions typically exist due to privacy concerns. Our aim is to design a privacy-preserving data-sharing mechanism that allows medical images to be stored as encoded and obfuscated representations in the public domain without revealing any useful or recoverable content from the images. In tandem, we aim to provide authorized users with compact private keys that could be used to reconstruct the corresponding images. METHOD:Our approach involves utilizing a neural auto-encoder. The convolutional filter outputs are passed through sparsifying transformations to produce multiple compact codes. Each code is responsible for reconstructing different attributes of the image. The key privacy-preserving element in this process is obfuscation through the use of specific pseudo-random noise. When applied to the codes, it becomes computationally infeasible for an attacker to guess the correct representation for all the codes, thereby preserving the privacy of the images. RESULTS:The proposed framework was implemented and evaluated using chest X-ray images for different medical image analysis tasks, including classification, segmentation, and texture analysis. Additionally, we thoroughly assessed the robustness of our method against various attacks using both supervised and unsupervised algorithms. CONCLUSION:This study provides a novel, optimized, and privacy-assured data-sharing mechanism for medical images, enabling multi-party sharing in a secure manner. While we have demonstrated its effectiveness with chest X-ray images, the mechanism can be utilized in other medical images modalities as well.

– To develop a robust and generalizable deep learning (DL) model gathering a massive and heterogenous dataset is crucial as the DL performances could be varied across different acquisition and reconstruction settings in the real clinical situation. Furthermore, sharing data with third parties is highly limited because of legal, ethical, security, and privacy issues. To address the aforementioned challenges, Federated learning (FL) allows one to train a DL model without sharing the data between different centers in a distributed and decentralized manner. In the current study, we developed deep FL-based models for intraprostatic lesion segmentation using FL approaches and compared their results by center based. Altogether 400 histologically proven prostate cancer patients with T2-weighted MRI images from eight different centers were enrolled. Dynamic data augmentation techniques for flipping left/right, elastic deformation, and random cropping was adopted. Two-stage cascaded U-Net consisting of modified 3D U-Ne, and Dual Attention 2D U-Net were implemented as the core of DL segmentation. MRI images and a prostate mask were used as input in this network. In addition, Federated Averaging (FedAvg) algorithm was implemented in this study. All evaluations were performed on 30% of test sets. In terms of dice and Jaccard coefficient, CeBa and achieved 0.77 ± 0.06 vs. 0.84 ± 0.05 (CI95%: 0.76 - 0.79 vs. 0.82 - 0.85) and 0.63 ± 0.09 vs. 0.72 ± 0.08 and (CI95%:0.61 - 0.65 vs. 0.70 - 0.74). We set out to develop a DL-based automated algorithm capable of segmenting intra-prostatic lesions using T2W MR images. Due to the complex structure and low-contrast nature of the intraprostatic lesion in T2W MR images, our proposed algorithm performed very well compared to manual segmentation across different centers. Our FL algorithms outperformed center base algorithms in which each center developed a model using their local dataset, which addresses data sharing between different centers.

BACKGROUND AND OBJECTIVE:Generalizable and trustworthy deep learning models for PET/CT image segmentation necessitates large diverse multi-institutional datasets. However, legal, ethical, and patient privacy issues challenge sharing of datasets between different centers. To overcome these challenges, we developed a federated learning (FL) framework for multi-institutional PET/CT image segmentation. METHODS:A dataset consisting of 328 FL (HN) cancer patients who underwent clinical PET/CT examinations gathered from six different centers was enrolled. A pure transformer network was implemented as fully core segmentation algorithms using dual channel PET/CT images. We evaluated different frameworks (single center-based, centralized baseline, as well as seven different FL algorithms) using 68 PET/CT images (20% of each center data). In particular, the implemented FL algorithms include clipping with the quantile estimator (ClQu), zeroing with the quantile estimator (ZeQu), federated averaging (FedAvg), lossy compression (LoCo), robust aggregation (RoAg), secure aggregation (SeAg), and Gaussian differentially private FedAvg with adaptive quantile clipping (GDP-AQuCl). RESULTS:The Dice coefficient was 0.80±0.11 for both centralized and SeAg FL algorithms. All FL approaches achieved centralized learning model performance with no statistically significant differences. Among the FL algorithms, SeAg and GDP-AQuCl performed better than the other techniques. However, there was no statistically significant difference. All algorithms, except the center-based approach, resulted in relative errors less than 5% for SUVmax and SUVmean for all FL and centralized methods. Centralized and FL algorithms significantly outperformed the single center-based baseline. CONCLUSIONS:The developed FL-based (with centralized method performance) algorithms exhibited promising performance for HN tumor segmentation from PET/CT images.

Success rate of clinical trials (CTs) is low, with the protocol design itself being considered a major risk factor. We aimed to investigate the use of deep learning methods to predict the risk of CTs based on their protocols. Considering protocol changes and their final status, a retrospective risk assignment method was proposed to label CTs according to low, medium, and high risk levels. Then, transformer and graph neural networks were designed and combined in an ensemble model to learn to infer the ternary risk categories. The ensemble model achieved robust performance (area under the receiving operator characteristic curve [AUROC] of 0.8453 [95% confidence interval: 0.8409-0.8495]), similar to the individual architectures but significantly out-performing a baseline based on bag-of-words features (0.7548 [0.7493-0.7603] AUROC). We demonstrate the potential of deep learning in predicting the risk of CTs from their protocols, paving the way for customized risk mitigation strategies during protocol design.

Purpose The generalizability and trustworthiness of deep learning (DL)–based algorithms depend on the size and heterogeneity of training datasets. However, because of patient privacy concerns and ethical and legal issues, sharing medical images between different centers is restricted. Our objective is to build a federated DL-based framework for PET image segmentation utilizing a multicentric dataset and to compare its performance with the centralized DL approach. Methods PET images from 405 head and neck cancer patients from 9 different centers formed the basis of this study. All tumors were segmented manually. PET images converted to SUV maps were resampled to isotropic voxels (3 × 3 × 3 mm 3 ) and then normalized. PET image subvolumes (12 × 12 × 12 cm 3 ) consisting of whole tumors and background were analyzed. Data from each center were divided into train/validation (80% of patients) and test sets (20% of patients). The modified R2U-Net was used as core DL model. A parallel federated DL model was developed and compared with the centralized approach where the data sets are pooled to one server. Segmentation metrics, including Dice similarity and Jaccard coefficients, percent relative errors (RE%) of SUV peak , SUV mean , SUV median , SUV max , metabolic tumor volume, and total lesion glycolysis were computed and compared with manual delineations. Results The performance of the centralized versus federated DL methods was nearly identical for segmentation metrics: Dice (0.84 ± 0.06 vs 0.84 ± 0.05) and Jaccard (0.73 ± 0.08 vs 0.73 ± 0.07). For quantitative PET parameters, we obtained comparable RE% for SUV mean (6.43% ± 4.72% vs 6.61% ± 5.42%), metabolic tumor volume (12.2% ± 16.2% vs 12.1% ± 15.89%), and total lesion glycolysis (6.93% ± 9.6% vs 7.07% ± 9.85%) and negligible RE% for SUV max and SUV peak . No significant differences in performance ( P > 0.05) between the 2 frameworks (centralized vs federated) were observed. Conclusion The developed federated DL model achieved comparable quantitative performance with respect to the centralized DL model. Federated DL models could provide robust and generalizable segmentation, while addressing patient privacy and legal and ethical issues in clinical data sharing.

To develop a robust and generalizable deep learning (DL) model gathering a massive and heterogenous dataset is crucial as the DL performances could be varied across different acquisition and reconstruction settings in the real clinical situation. Furthermore, sharing data with third parties is highly limited because of legal, ethical, security, and privacy issues. To address the aforementioned challenges, Federated learning (FL) allows one to train a DL model without sharing the data between different centers in a distributed and decentralized manner. In the current study, we developed deep FL-based models for intraprostatic lesion segmentation using FL approaches and compared their results by center based. Altogether 400 histologically proven prostate cancer patients with T2-weighted MRI images from eight different centers were enrolled. Dynamic data augmentation techniques for flipping left/right, elastic deformation, and random cropping was adopted. Two-stage cascaded U-Net consisting of modified 3D U-Ne, and Dual Attention 2D U-Net were implemented as the core of DL segmentation. MRI images and a prostate mask were used as input in this network. In addition, Federated Averaging (FedAvg) algorithm was implemented in this study. All evaluations were performed on 30% of test sets. In terms of dice and Jaccard coefficient, CeBa and achieved 0.77 ± 0.06 vs. 0.84 ± 0.05 (CI95%: 0.76 - 0.79 vs. 0.82 - 0.85) and 0.63 ± 0.09 vs. 0.72 ± 0.08 and (CI95%:0.61 - 0.65 vs. 0.70 - 0.74). We set out to develop a DL-based automated algorithm capable of segmenting intra-prostatic lesions using T2W MR images. Due to the complex structure and low-contrast nature of the intraprostatic lesion in T2W MR images, our proposed algorithm performed very well compared to manual segmentation across different centers. Our FL algorithms outperformed center base algorithms in which each center developed a model using their local dataset, which addresses data sharing between different centers.

In this paper, we consider privacy-preserving compressed image sharing, where the goal is to release compressed data whilst satisfying some privacy/secrecy constraints yet ensuring image reconstruction with a defined fidelity. The privacy-preserving compressed image sharing is addressed using a machine learning framework based on an information bottleneck with a shared secret key for authorized users. In contrast, an adversary observing the protected compressed representation tries to either reconstruct the data or deduce some privacy-sensitive attributes such as gender, age, etc. The inference task on the adversary’s side is performed without the knowledge of the shared secret key and is based on an adversarial mutual information maximization between the privacy-protected compressed representation and targeted attributes. The proposed framework is experimentally validated on the CelebA dataset.

As the world's population continues to expand, maritime transport is critical to ensure economic growth. To improve security and safety of maritime transportation, the Automatic Identification System (AIS) collects real-time data about vessels and their positions. While a large portion of the AIS data is provided via an automatic tracking system, some key fields, such as destination and draught, are entered manually by the ship navigator and are thus prone to errors. To support decision making in maritime industries, in this paper we propose a data-driven vessel destination prediction algorithm based on heterogeneous graph and machine learning models. We design the task as a multi-class classification problem, where the destination port is the category to be predicted given the vessel and origin information. Then, we use a link prediction model in a weighted heterogeneous graph to predict the vessel destination. Experimental comparison against baseline methods, such as logistic regression and k-nearest neighbors, showed that our model provides a robust performance, outperforming the baseline algorithms by 9% and 33% in terms of accuracy and F1-score, respectively. Thus, heterogeneous graph models provide a powerful alternative to predict port destination, and could support enhancing AIS data quality and better decision making in maritime transportation industries.

A recent trend in health-related machine learning proposes the use of Graph Neural Networks (GNN’s) to model biomedical data. This is justified due to the complexity of healthcare data and the modelling power of graph abstractions. Thus, GNN’s emerge as the natural choice to learn from increasing amounts of healthcare data. While formulating the problem, however, there are usually multiple design choices and decisions that can affect the final performance. In this work, we focus on Clinical Trial (CT) protocols consisting of hierarchical documents, containing free text as well as medical codes and terms, and design a classifier to predict each CT protocol termination risk as “low” or “high”. We show that while using GNN’s to solve this classification task is very successful, the way the graph is constructed is also of importance and one can benefit from making a priori useful information more explicit. While a natural choice is to consider each CT protocol as an independent graph and pose the problem as a graph classification, consistent performance improvements can be achieved by considering them as super-nodes in one unified graph and connecting them according to some metadata, like similar medical condition or intervention, and finally approaching the problem as a node classification task rather than graph classification. We validate this hypothesis experimentally on a large-scale manually labeled CT database. This provides useful insights on the flexibility of graph-based modeling for machine learning in the healthcare domain.

A small dataset commonly affects generalization, robustness, and overall performance of deep neural networks (DNNs) in medical imaging research. Since gathering large clinical databases is always difficult, we proposed an analytical method for producing a large realistic/diverse dataset. Clinical brain PET/CT/MR images including full-dose (FD), low-dose (LD) corresponding to only 5 % of events acquired in the FD scan, non-attenuated correction (NAC) and CT-based measured attenuation correction (MAC) PET images, CT images and T1 and T2 MR sequences of 35 patients were included. All images were registered to the Montreal Neurological Institute (MNI) template. Laplacian blending was used to make a natural presentation using information in the frequency domain of images from two separate patients, as well as the blending mask. This classical technique from the computer vision and image processing communities is still widely used and unlike modern DNNs, does not require the availability of training data. A modified ResNet DNN was implemented to evaluate four image-to-image translation tasks, including LD to FD, LD+MR to FD, NAC to MAC, and MRI to CT, with and without using the synthesized images. Quantitative analysis using established metrics, including the peak signal-to-noise ratio (PSNR), structural similarity index metric (SSIM), and joint histogram analysis was performed for quantitative evaluation. The quantitative comparison between the registered small dataset containing 35 patients and the large dataset containing 350 synthesized plus 35 real dataset demonstrated improvement of the RMSE and SSIM by 29% and 8% for LD to FD, 40% and 7% for LD+MRI to FD, 16% and 8% for NAC to MAC, and 24% and 11% for MRI to CT mapping task, respectively. The qualitative/quantitative analysis demonstrated that the proposed model improved the performance of all four DNN models through producing images of higher quality and lower quantitative bias and variance compared to reference images.

Attenuation correction and scatter compensation (AC/SC) are two main steps toward quantitative PET imaging, which remain challenging in PET-only and PET/MRI systems. These can be effectively tackled via deep learning (DL) methods. However, trustworthy, and generalizable DL models commonly require well-curated, heterogeneous, and large datasets from multiple clinical centers. At the same time, owing to legal/ethical issues and privacy concerns, forming a large collective, centralized dataset poses significant challenges. In this work, we aimed to develop a DL-based model in a multicenter setting without direct sharing of data using federated learning (FL) for AC/SC of PET images. Non-attenuation/scatter corrected and CT-based attenuation/scatter corrected (CT-ASC) 18F-FDG PET images of 300 patients were enrolled in this study. The dataset consisted of 6 different centers, each with 50 patients, with scanner, image acquisition, and reconstruction protocols varying across the centers. CT-based ASC PET images served as the standard reference. All images were reviewed to include high-quality and artifact-free PET images. Both corrected and uncorrected PET images were converted to standardized uptake values (SUVs). We used a modified nested U-Net utilizing residual U-block in a U-shape architecture. We evaluated two FL models, namely sequential (FL-SQ) and parallel (FL-PL) and compared their performance with the baseline centralized (CZ) learning model wherein the data were pooled to one server, as well as center-based (CB) models where for each center the model was built and evaluated separately. Data from each center were divided to contribute to training (30 patients), validation (10 patients), and test sets (10 patients). Final evaluations and reports were performed on 60 patients (10 patients from each center). In terms of percent SUV absolute relative error (ARE

PURPOSE:Reducing the injected activity and/or the scanning time is a desirable goal to minimize radiation exposure and maximize patients' comfort. To achieve this goal, we developed a deep neural network (DNN) model for synthesizing full-dose (FD) time-of-flight (TOF) bin sinograms from their corresponding fast/low-dose (LD) TOF bin sinograms. METHODS:Clinical brain PET/CT raw data of 140 normal and abnormal patients were employed to create LD and FD TOF bin sinograms. The LD TOF sinograms were created through 5% undersampling of FD list-mode PET data. The TOF sinograms were split into seven time bins (0, ±1, ±2, ±3). Residual network (ResNet) algorithms were trained separately to generate FD bins from LD bins. An extra ResNet model was trained to synthesize FD images from LD images to compare the performance of DNN in sinogram space (SS) vs implementation in image space (IS). Comprehensive quantitative and statistical analysis was performed to assess the performance of the proposed model using established quantitative metrics, including the peak signal-to-noise ratio (PSNR), structural similarity index metric (SSIM) region-wise standardized uptake value (SUV) bias and statistical analysis for 83 brain regions. RESULTS:SSIM and PSNR values of 0.97 ± 0.01, 0.98 ± 0.01 and 33.70 ± 0.32, 39.36 ± 0.21 were obtained for IS and SS, respectively, compared to 0.86 ± 0.02and 31.12 ± 0.22 for reference LD images. The absolute average SUV bias was 0.96 ± 0.95% and 1.40 ± 0.72% for SS and IS implementations, respectively. The joint histogram analysis revealed the lowest mean square error (MSE) and highest correlation (R2 = 0.99, MSE = 0.019) was achieved by SS compared to IS (R2 = 0.97, MSE= 0.028). The Bland & Altman analysis showed that the lowest SUV bias (-0.4%) and minimum variance (95% CI: -2.6%, +1.9%) were achieved by SS images. The voxel-wise t-test analysis revealed the presence of voxels with statistically significantly lower values in LD, IS, and SS images compared to FD images respectively. CONCLUSION:The results demonstrated that images reconstructed from the predicted TOF FD sinograms using the SS approach led to higher image quality and lower bias compared to images predicted from LD images.

We consider the hierarchical representation of documents as graphs and use geometric deep learning to classify them into different categories. While graph neural networks can efficiently handle the variable structure of hierarchical documents using the permutation invariant message passing operations, we show that we can gain extra performance improvements using our proposed selective graph pooling operation that arises from the fact that some parts of the hierarchy are invariable across different documents. We applied our model to classify clinical trial (CT) protocols into completed and terminated categories. We use bag-of-words based, as well as pre-trained transformer-based embeddings to featurize the graph nodes, achieving f1-scores around 0.85 on a publicly available large scale CT registry of around 360K protocols. We further demonstrate how the selective pooling can add insights into the CT termination status prediction. We make the source code and dataset splits accessible.