Studies in health technology and informatics最新文献

This article explores the transition from a traditional histopathological examination system to an innovative platform using artificial intelligence (AI) for breast cancer detection from histopathological images in Burkina Faso. The existing system is analyzed in detail, highlighting the steps of querying, sample preparation, analysis by the pathologist, and validation by the physician. From this analysis, the needs and challenges are identified, emphasizing the opportunities for AI to improve the efficiency and accuracy of the diagnosis. The design of the AI platform is then presented, including data collection, AI model development, and its integration into existing processes. Finally, the expected results and implications for improving healthcare in Burkina Faso are discussed, highlighting the potential benefits and challenges to overcome for the successful adoption of this promising technology.

Common Data Models (CDM) are developed to solve integration problems that arise in the secondary use of health data. The OMOP CDM is such a model that is mainly used for healthcare data, so this paper examines whether it is also suitable for mapping research data. An exemplary research dataset is mapped to the model and the model is tested for suitability. For this purpose, an ETL process is first designed with the OHDSI tools and finally implemented with Talend Open Studio for Data Integration. The data quality is checked, and the mapping and the model, together with the tools, are evaluated. Overall, all but three data fields from the source dataset could be mapped to the OMOP model, so that the model's suitability for research data can be confirmed.

Polypharmacy (PP) and hyperpolypharmacy (HPP), are prevalent among cancer patients and are associated with an increased risk of drug-drug interactions (DDI) and potentially inappropriate medications (PIM). This study aimed to characterize PP, HPP, DDI, and PIM in patients with hematological malignancies hospitalized for hematopoietic stem cell transplantation (HSCT) by introducing a novel metric: cumulative drug exposure. Clinical data warehouse (CDW) records were employed to develop algorithms that quantified patients' cumulative exposure to these prescribing determinants during hospitalization. This entailed determining the number of days during the hospital stay when the patient was exposed to PP, HPP, PIM and/or DDI. For PIM and DDI, the number of PIMs or DDIs administered per day was taken into account in this calculation. Among 339 HSCT patients, PP and HPP were highly prevalent (over 67% of HSCT patients), almost all patients experienced DDI (over 98% of HSCT patients) and almost all elderly patients were exposed to PIM (over 98% of HSCT patients). Cumulative drug exposure differed between allogeneic and autologous HSCT patients, with allogeneic patients being more exposed to HPP (28.5 days vs 4.7 days for autologous HSCT patients) and DDI (255.6 days vs 58.4 for autologous HSCT patients). This study proposes a novel approach to assessing the impact of prescribing determinants on patient outcomes and provides insights for future research into the association between drug exposure and adverse events. Indeed, the use of cumulative drug exposure as a metric provides a comprehensive view of patient exposure throughout hospitalization, thereby enhancing understanding of the impact of prescribing practices on clinical outcomes.

The existing pill dispenser systems help elderly people to improve their quality of life, and medication adherence. But these systems lack interactive capabilities with caregivers, a crucial element in comprehensive home care management. The suggested CareConnect aims to bridge this gap by introducing a Telecare extension that not only manages medication adherence but also facilitates interaction between the patient and their caregivers. The system is described in terms of this new approach, the functions, hardware and software. The operation of the system is briefly described. A discussion about the advantages of CareConnect system and the future development directions is finally done as a conclusion.

In Generative Artificial Intelligence (AI), Large Language Models (LLMs) like GPT-4, Gemini, Claude, and Llama, significantly impact healthcare by aiding in patient care, medical research, and administrative tasks. AI-powered chatbots offer real-time responses and manage chronic diseases, improving patient outcomes and operational efficiency. However, these models pose security and ethical challenges, necessitating robust data privacy, adversarial training, and ethical guidelines. This paper proposes a secure, ethical pipeline for deploying AI healthcare chatbots, integrating advanced privacy-preserving techniques and continuous security assessments to enhance data privacy, resilience, and user trust.

The most well-established risk factor for lung cancer (LC) is smoking, responsible for approximately 85% of cases. The Lung Cancer Risk Assessment Tool (LCRAT) is a key advancement in this field, which predicts individual risk based on factors like smoking habits, demographic details, personal and family medical history, and environmental exposures. This paper proposes a model with fewer features that improves state of the art performance, using a simplified stacking ensemble, making it more accessible and easier to implement in routine healthcare practice. The data used in this work were derived from two cohorts in the United States: The National Lung Screening Trial (NLST) and the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. Both our model and LCRAT achieve an AUC of 0.799 and 0.782 on test respectively. In terms of percentage of positives, in the 50% of the population, both detect 0.766 and 0.754 of the cases. The ensemble of different survival models enhances robustness by mitigating the weakness of individual models and directly impacts the efficiency of the model, increasing the efficiency and generalizability.

The ability to recognize anatomical landmarks, microsurgical instruments, and complex scenes and events in a surgical wound using computer vision presents new opportunities for studying microsurgery effectiveness. In this study, we aimed to develop an artificial intelligence-based solution for detecting, segmenting, and tracking microinstruments using a neurosurgical microscope. We have developed a technique to process videos from microscope camera, which involves creating a segmentation mask for the instrument and subsequently tracking it. We compared two segmentation approaches: (1) semantic segmentation using Visual Transformers (pre-trained domain-specific EndoViT model), enhanced with tracking as described by Cheng Y. et al. (our proposed approach), and (2) instance segmentation with tracking based on the YOLOv8l-seg architecture. We conducted experiments using the CholecSeg8k dataset and our proprietary set of neurosurgical videos (PSNV) from microscope. Our approach with tracking outperformed YOLOv8l-seg-based solutions and EndoViT model with no tracking on both CholecSeg8k (mean IoT = 0.8158, mean Dice = 0.8657) and PSNV (mean IoT = 0.7196, mean Dice = 0.8202) datasets. Our experiments with identifying neurosurgical instruments in a microscope's field of view showcase the high quality of these technologies and their potential for valuable applications.

Natural Language Processing (NLP) has shown promise in fields like radiology for converting unstructured into structured data, but acquiring suitable datasets poses several challenges, including privacy concerns. Specifically, we aim to utilize Large Language Models (LLMs) to extract medical information from dialogues between ambulance staff and patients to populate emergency protocol forms. However, we currently lack dialogues with known content that can serve as a gold standard for an evaluation. We designed a pipeline using the quantized LLM "Zephyr-7b-beta" for initial dialogue generation, followed by refinement and translation using OpenAI's GPT-4 Turbo. The MIMIC-IV database provided relevant medical data. The evaluation involved accuracy assessment via Retrieval-Augmented Generation (RAG) and sentiment analysis using multilingual models. Initial results showed a high accuracy of 94% with "Zephyr-7b-beta," slightly decreasing to 87% after refinement with GPT-4 Turbo. Sentiment analysis indicated a qualitative shift towards more positive sentiment post-refinement. These findings highlight the potential and challenges of using LLMs for generating synthetic medical dialogues, informing future NLP system development in healthcare.

Incorporating digital medicine into medical education equips students for the evolving landscape of healthcare. This study aimed to assess a digital medicine course developed at Bielefeld University by evaluating student attainment of learning outcomes outlined by Foadi et al. In the course, the students designed a digital application for various medical conditions, taking into account interdisciplinary factors. The course took place in 2023 with medical students who attended the course due to the focus of their studies. In a pilot study, the progress of ten participants was assessed using a pre-post survey design. Results revealed substantial improvement in students' achievement of learning outcomes post-course (median = 2, IQR 1-2) compared to pre-course (median = 3, IQR 3-4), suggesting the course's efficacy in effectively teaching digital medicine.

GNU Health, an open-source clinical information system, offers a comprehensive solution for managing health records, hospital information, and laboratory data. Despite its robust functionality and cost-effective nature, GNU Health remains underutilized in the European healthcare context. This paper explores the potential benefits of implementing GNU Health in European healthcare systems, emphasizing its capacity for customization, integration, and scalability. We also examine the barriers to its widespread adoption, including regulatory challenges, interoperability issues, and resistance to change from established proprietary systems. Through one case study and expert interviews, we provide insights into why these obstacles can hardly be overcome.