Luis Manuel Nobre de Brito Elvas

• Ferreira, Joao C. & Elvas, Luìs B. (2026). Time-Aware Machine Learning for Prospective Tachycardia Prediction in Cardiovascular Disease. Intelligence-Based Medicine. doi: 10.1016/j.ibmed.2026.100482. Full text in Research Archive Show summary

  Tachycardia can accompany acute deterioration in hospitalized patients with cardiovascular disease, yet most machine-learning studies emphasize detection rather than prospective prediction. Using routinely collected physiological monitoring data from a hospital-wide database of 512,764 patients, we identified 968 patients with myocardial infarction (MI; n = 260), pulmonary thromboembolism (PTE; n = 48), or aortic stenosis (AS; n = 660) and developed models for prospective tachycardia prediction. Tachycardia was defined as heart rate >100 bpm. Historical physiological features were separated temporally from the future target, population-level autocorrelation analysis informed lag selection, and Random Forest (RF), Naïve Bayes (NB), and a feedforward neural network (NN) were evaluated using patient-blocked, time-aware validation. At the selected 90-step lag configuration, RF showed the strongest performance, including an F1-score of 0.89 ± 0.02, PR-AUC of 0.72 ± 0.03, and AUROC of 0.93 ± 0.02 for PTE. These findings provide internal feasibility evidence for time-aware tachycardia risk prediction from routine hospital monitoring data; external and prospective validation is required before clinical deployment.

• Ferreira, Joao C. & Elvas, Luìs B. (2026). Machine learning-based early detection of abnormal heart rate in critically ill patients: a real-world clinical dataset study with automated alert system integration. Frontiers in Artificial Intelligence. 9. doi: 10.3389/frai.2026.1879609. Full text in Research Archive Show summary

  Introduction: Cardiovascular diseases (CVD) are a leading global health challenge, particularly in resource-constrained settings where delayed diagnosis worsens outcomes. Abnormal heart rate (tachycardia/bradycardia) is a modifiable, independent risk factor that benefits from early, automated detection.Methods: We used a large real-world dataset from Hospital Santa Maria, Lisbon (23,122 patients, January 2019–November 2021, spanning the COVID-19 period). Two operational outcome definitions were evaluated: (i) an interquartile-range (IQR)-based, patient-specific definition, used to compare four classifiers (Logistic Regression, K-Nearest Neighbors, Random Forest, Naive Bayes), and (ii) standard clinical thresholds for tachycardia/bradycardia, applied to the best-performing model. Following CRISP-DM, models were trained and evaluated with patient-grouped, walk-forward time-series cross-validation to prevent patient-level data leakage.Results: Logistic Regression was the best-performing model. Under the IQR-based definition: AUROC 0.78, recall 85.4%, precision 66.1%, F1-score 74.5%. Under the clinically standard threshold-based definition: AUROC 0.81, recall 91.7%, precision 68.8%, F1-score 78.6%. Expanded metrics (event prevalence, false-positive rate, alert burden, calibration, confidence intervals) are reported. A prototype real-time e-mail alert system was demonstrated as a feasibility proof-of-concept and reviewed informally by two cardiovascular specialists.Discussion: Recall and AUROC alone do not establish clinical utility. Key limitations include the single-centre design, substantial missingness in ethnicity/diagnosis variables, the small clinician-evaluation sample (n = 2), and pandemic-period data collection. This work is presented as a methodologically strengthened proof-of-concept, with prospective, multi-site, multi-clinician external validation identified as the necessary next step.

• Fazendas, Paula; Bairros, Rita; Elvas, Luìs B.; Brochado, Liliana; Ferreira, Joao C. & Gomes, Rita [Show all 13 contributors for this article] (2026). Artificial intelligence–based calcium scoring using 3D transesophageal echocardiography in aortic stenosis: a pilot study. Journal of Cardiovascular Imaging. ISSN 2586-7210. 34(1). doi: 10.1186/s44348-026-00080-x. Full text in Research Archive Show summary

  Abstract Background Aortic valve calcium scoring by computed tomography (CT) is an established method for assessing aortic stenosis severity but is limited by radiation exposure and availability. Artificial intelligence (AI)-based calcium detection using transthoracic echocardiography has shown promise but depends on acoustic window quality. Transesophageal echocardiography (TEE), particularly 3D TEE, may overcome these limitations by providing improved visualization without radiation. The objective of this study is to evaluate the feasibility of AI-based quantification of aortic valve calcium using 3D TEE. Methods In this prospective pilot study, 23 patients (median age, 76 years; 56.5% male) with moderate or severe aortic stenosis underwent 3D TEE and CT. Multiplanar reconstruction generated 1.5-mm diastolic short-axis slices. A computer vision–based model identified calcium-related speckles. An automated TEE calcium score was derived from the sum of calcium pixels across 11 frames per patient, which was compared with the CT Agatston score. Results The TEE calcium score showed a significant positive correlation with CT Agatston scores (r = 0.65, P < 0.001). Receiver operating characteristic analysis yielded an area under the curve of 0.87 (95% confidence interval, 0.69–1.00) for identifying severe calcification (cutoff, 68,813 pixels; sensitivity, 89.5%; specificity, 75.0%). Conclusions AI-based calcium quantification using 3D TEE is feasible and correlates with CT-derived scores. This radiation-free approach may provide a promising alternative for assessing aortic valve calcification.

• Elvas, Luìs B. & Carvalho, Carolina (2026). Domain-specific versus general large language models: a review and empirical benchmark in real medical texts. Scientific Reports. 16. doi: 10.1038/s41598-026-62884-7. Full text in Research Archive Show summary

  Abstract The rapid adoption of Large Language Models (LLMs) has led to their widespread use as a general-purpose, fast-deployment solution across a broad range of tasks. In healthcare settings, this trend has accelerated the replacement of task-specific models with large generative architectures, often without an assessment of their efficiency or suitability for structured information extraction. While transformer-based encoder models have long been established as reliable solutions for clinical Named Entity Recognition (NER) and Information Extraction (IE), the increasing reliance on LLMs raises concerns regarding computational cost, scalability, energy consumption, and practical deployment in resource- and privacy-constrained environments. This study aimed to critically evaluate the assumption that LLMs constitute a universally superior solution for clinical NER and IE, by systematically comparing encoder-based Language Models (LMs), and LLMs in terms of extraction performance, efficiency, deployment feasibility, and environmental impact. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines, a scoping literature review was conducted using Scopus and Web of Science, covering peer-reviewed studies published between 2022 and 2025. In addition, a proof-of-concept case study was performed on European Portuguese clinical notes, comparing a domain-adapted encoder model (MediAlbertina-1.5B) with quantized Llama-based LLMs executed locally under identical hardware conditions. Models were evaluated using micro-averaged precision, recall, and F1-score, together with runtime, estimated energy consumption, and CO₂ emissions. The literature review showed that LLMs and transformer-based architectures were the most frequently used even thought encoder-based and domain-adapted models often perform strongly in clinical NER, particularly when recall, structured output, and computational efficiency are important. The empirical case study showed a similar pattern: MediAlbertina-1.5B achieved the highest performance (micro-F1 = 0.430), whereas all evaluated Llama variants obtained markedly lower F1-scores (≤ 0.123) due to systematic under-extraction. Despite their lower performance, the Llama models incurred 23–95 × longer runtimes and one to two orders of magnitude higher estimated energy consumption and CO₂ emissions under the tested local CPU-only setup. In this proof-of-concept benchmark on 20 European Portuguese clinical reports, the domain-adapted MediAlbertina token-classification model achieved higher recall and micro-F1 than the evaluated locally executed, quantized Llama models. The Llama models were also substantially slower and less sustainable, with higher estimated energy use and CO₂ emissions under the tested hardware and prompting configuration. These findings suggest that domain-adapted encoder models may be preferable for similar structured clinical NER settings, although larger and externally validated benchmarks are needed before generalizing to other languages, clinical corpora, model families, or deployment environments.

• Juma, Alexandre; Elvas, Luìs B.; Ferreira, Joao C. & Nunes, Luis (2026). Mobility effect on city pollution: A case study. Journal of Ambient Intelligence and Smart Environments. ISSN 1876-1364. 18(2), p. 234–248. doi: 10.1177/18761364251396640. Full text in Research Archive Show summary

  The present work reports the impacts on urban mobility and air quality in Lisbon, Portugal, of the imposed restrictions to curb the transmission of SARS-CoV-2 virus, which causes COVID-19 disease. We performed a data-driven approach over Lisbon Smart cities data, collected from several sources, such as traffic and pollution. During the first Portuguese emergency period (18-03-2020 to 03-05-2020) the sharp reductions in anthropogenic activities, most importantly road traffic, resulted in generally reduced criteria air pollutant concentration compared to an homologous baseline from 2013–2019 measured in the six air quality monitoring stations throughout the city. The most negatively impacted air pollutants were NO2, with a reduction of 54.35% in traffic stations and 28.62% in background stations. Google mobility indicator for local commerce was found to be the main anthropogenic activity indicator for Lisbon, with a moderate and positive correlation with NO2 concentration (r=+0.54). A regressor ML pipeline was trained to predict NO2 concentration with the available anthropogenic activity, weather, and air pollutant inputs from March/2020 to March/2021, achieving R2 = 0.925 on the test set.

• Elvas, Luìs B.; Oliveira, Bruno; Mira, Miguel & Ferreira, Joao C. (2025). AI-Enabled Knowledge Creation: A Unified Database Approach in a Multinational Hospital Network. Proceedings of the Annual Hawaii International Conference on System Sciences (HICSS). ISSN 1530-1605. 58, p. 3453–3462. doi: 10.24251/hicss.2025.416. Full text in Research Archive Show summary

  This study outlines an approach for integrating health data from 72 hospitals in 13 countries, focusing on COVID-19 patients with cardiovascular issues. The goal is to combine data within a European project, prioritizing data privacy and utilizing an Extract, Transform, and Load (ETL) architecture for efficient data management. Machine learning, including AI, is applied to predict anomalies in patient data, enhancing the ETL process's capability to support such algorithms. An alert system is established to flag potential outliers for swift medical attention. Challenges such as interoperability and privacy are addressed, and the study evaluates the ETL and AI methods against key performance metrics, confirming their effectiveness. The unified database allows for benchmarking and sharing best practices across hospitals, improving healthcare quality. The paper contributes by detailing ETL development challenges in healthcare and showcasing the benefits of a centralized data repository for healthcare management, particularly through a machine learning algorithm designed to predict abnormal patient values, thereby aiding healthcare professionals in decision-making and improving patient care.

• Phuyal, Sudip; Elvas, Luìs B.; Ferreira, Joao C. & Bista, Rabindra (2025). Blockchain technology in healthcare : unifying patient medical records : a survey. In Sakalauskas, Virgilijus; Bajaj, Anu; Abraham, Ajith; Madhavi, K. Reddy & Mishra, Pooja Manghirmalani (Ed.), Bio-Inspired Computing : Proceedings of the 14th International Conference on Innovations in Bio-Inspired Computing and Applications (IBICA 2023) held during December 14-15, 2023, Volume 4: Machine Learning Applications. Springer. ISSN 9783031789465. p. 319–332. doi: 10.1007/978-3-031-78946-5_31. Full text in Research Archive Show summary

  In the era of digital transformation, healthcare faces a critical challenge in managing and centralizing patient medical records efficiently and securely. Traditional data management systems, including paper-based records and fragmented electronic health records (EHR) systems, have demonstrated limitations in terms of data security, interoperability, and patient privacy. Blockchain technology has become known as a possible solution to all of these problems, one that has an opportunity to completely transform the administration of healthcare data management. This paper presents a comprehensive exploration of the application of blockchain technology in healthcare, specifically focusing on centralizing patient medical records. Through a literature review and analysis of real-world initiatives, we examine the potential of blockchain to address the shortcomings of existing systems. Blockchain’s key characteristics, such as decentralization, data immutability, and cryptographic security, offer a unique framework for enhancing data security, improving interoperability among healthcare providers, and empowering patients to take control of their health information. The literature review highlights previous research and case studies that illustrate how blockchain has been successfully implemented in healthcare settings. We discuss notable initiatives like MedRec and Medicalchain, showcasing their impact on data security, patient care coordination, and data sharing within the healthcare ecosystem. Consequently, this article offers insightful information about how blockchain technology could revolutionize the handling of healthcare data. It underscores the importance of blockchain as a solution to the challenges associated with centralizing patient medical records, ultimately contributing to improved healthcare outcomes, enhanced patient experiences, and the future evolution of healthcare delivery. This research serves as a resource for healthcare professionals, researchers, and policymakers seeking innovative solutions to enhance data management and patient-centric care in the everevolving healthcare landscape. Blockchain technology adoption in the healthcare industry has the ability to completely transform security of data and accessibility, resulting in a more effective and patient-centered healthcare system.

• Phuyal, Sudip; Elvas, Luìs B.; Ferreira, Joao C. & Bista, Rabindra (2025). Wearable devices for long-term care : survey and opportunities. In Sakalauskas, Virgilijus; Bajaj, Anu; Abraham, Ajith; Madhavi, K. Reddy & Mishra, Pooja Manghirmalani (Ed.), Bio-Inspired Computing : Proceedings of the 14th International Conference on Innovations in Bio-Inspired Computing and Applications (IBICA 2023) held during December 14-15, 2023, Volume 4: Machine Learning Applications. Springer. ISSN 9783031789465. p. 119–128. doi: 10.1007/978-3-031-78946-5_11. Full text in Research Archive Show summary

  The incorporation of wearable devices within long-term care represents a groundbreaking advancement in healthcare. These devices, leveraging sensor and intelligent technologies, possess the potential to revolutionize the supervision and handling of chronic illnesses, enriching the lives of individuals reliant on extended care. This paper conducts an exhaustive examination of wearable devices in the long-term care context, spotlighting their wide-ranging applications. It encompasses devices tailored for monitoring vital signs, mobility, medication adherence, cognitive health, and other critical facets of long-term care. Through an in-depth exploration of their capabilities and limitations, it illuminates their capacity to ease the strain on healthcare systems and enhance patient well-being. Moreover, this paper investigates the hurdles related to wearable device adoption in long-term care, addressing concerns like data security, user acceptance, and interoperability. By tackling these challenges, it endeavors to chart a path for the seamless integration of wearable technologies. Drawing insights from current research and industry trends, it identifies emerging prospects such as personalized care plans, remote monitoring, and data-driven healthcare interventions, fostering collaboration among healthcare stakeholders. By harnessing the full potential of wearable technologies, the quality of life of individuals in long-term care can be improved and also, it can contribute to achieve more efficient and sustainable healthcare systems.

• Elvas, Luìs B.; Santos, Rafaela & Ferreira, Joao C. (2025). Fine-tuning of language models for automated structuring of medical exam reports to improve patient screening and analysis. Scientific Reports. 15. doi: 10.1038/s41598-025-05695-6. Full text in Research Archive Show summary

  The analysis of medical imaging reports is labour-intensive but crucial for accurate diagnosis and effective patient screening. Often presented as unstructured text, these reports require systematic organisation for efficient interpretation. This study applies Natural Language Processing (NLP) techniques tailored for European Portuguese to automate the analysis of cardiology reports, streamlining patient screening. Using a methodology involving tokenization, part-of-speech tagging and manual annotation, the MediAlbertina PT-PT language model was fine-tuned, achieving 96.13% accuracy in entity recognition. The system enables rapid identification of conditions such as aortic stenosis through an interactive interface, substantially reducing the time and effort required for manual review. It also facilitates patient monitoring and disease quantification, optimising healthcare resource allocation. This research highlights the potential of NLP tools in Portuguese healthcare contexts, demonstrating their applicability to medical report analysis and their broader relevance in improving efficiency and decision-making in diverse clinical environments.

• Elvas, Luìs B.; Almeida, Ana & Ferreira, Joao C. (2025). Natural language processing in medical text processing : a scoping literature review. International Journal of Medical Informatics. ISSN 1386-5056. 204(December). doi: 10.1016/j.ijmedinf.2025.106049. Full text in Research Archive Show summary

  Background: The exponential growth of digitized medical data has created significant challenges for healthcare professionals, as medical documentation transitions from simple text records to complex, multi-dimensional data structures. Natural Language Processing (NLP), particularly Named Entity Recognition (NER), has emerged as a crucial tool for extracting and categorizing critical information from clinical texts. The development of transformer-based models like BERT and the ability to fine-tune pre-trained AI models have revolutionized the field, offering unprecedented opportunities to enhance the efficient and precise interpretation of medical data across diverse languages and healthcare contexts. Objective: This literature review aimed to analyze recent NLP approaches for medical text processing, examining techniques, performance metrics, and advancements across different languages and healthcare contexts. Method: Following the Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) methodology, a scoping search was conducted in Scopus and PubMed databases, focusing on studies published between 2019–2024. The review included studies on language model fine-tuning and information extraction in healthcare, with a specific search query designed to capture relevant NLP techniques. Results: Of 67 initial records, 31 studies were ultimately included. Bidirectional Encoder Representations from Transformers (BERT)-based approaches, neural networks, and CRF/LSTM techniques dominated, consistently achieving F1-scores above 85 %. The studies covered multiple languages, with 51.5 % in English, 27.3 % in Chinese, and smaller representations in Italian, German, and Spanish. Hybrid approaches and techniques addressing data privacy and limited labeled data were notably prevalent. Conclusions: The review revealed that modern NLP techniques, particularly BERT-based models and hybrid approaches, show significant promise in medical text processing across different languages. While challenges remain in cross-lingual adaptation and data availability, these technologies demonstrate potential to enhance medical data interpretation and analysis.

• Ferreira, Joao C.; Elvas, Luìs B.; Correia, Ricardo & Mascarenhas, Miguel (2025). Empowering Health Professionals with Digital Skills to Improve Patient Care and Daily Workflows. Healthcare. 13(3). doi: 10.3390/healthcare13030329. Full text in Research Archive Show summary

  The increasing digitalisation of healthcare has created a pressing need for health professionals to develop robust digital skills. This paper explores the imperative of equipping health professionals with the necessary digital proficiency to enhance their daily workflows and improve patient care. The expanding use of digital technologies, including electronic health records, telehealth, and artificial intelligence, has transformed the healthcare landscape. However, the adoption of these technologies has been hindered by barriers, such as a lack of interoperability and hesitancy among healthcare providers. To address these challenges, this paper argues that digital skill development must be a core component of healthcare education and professional training. Medical schools and healthcare organisations must prioritise the integration of digital health curricula and continuous learning opportunities to ensure that the next generation of healthcare providers is well-equipped to navigate the digital healthcare ecosystem. Additionally, this paper highlights the importance of fostering a culture of digital innovation and collaboration within healthcare settings. By empowering health professionals to actively participate in the development and testing of new digital health applications, the industry can unlock the full potential of digital technologies to enhance daily workflows and improve patient outcomes.

• Elvas, Luìs B.; Almeida, Ana & Ferreira, Joao C. (2025). The Role of AI in Cardiovascular Event Monitoring and Early Detection: Scoping Literature Review. JMIR Medical Informatics. 13. doi: 10.2196/64349. Full text in Research Archive Show summary

  Background: Artificial intelligence (AI) has shown exponential growth and advancements, revolutionizing various fields, including health care. However, domain adaptation remains a significant challenge, as machine learning (ML) models often need to be applied across different health care settings with varying patient demographics and practices. This issue is critical for ensuring effective and equitable AI deployment. Cardiovascular diseases (CVDs), the leading cause of global mortality with 17.9 million annual deaths, encompass conditions like coronary heart disease and hypertension. The increasing availability of medical data, coupled with AI advancements, offers new opportunities for early detection and intervention in cardiovascular events, leveraging AI’s capacity to analyze complex datasets and uncover critical patterns. Objective: This review aims to examine AI methodologies combined with medical data to advance the intelligent monitoring and detection of CVDs, identifying areas for further research to enhance patient outcomes and support early interventions. Methods: This review follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology to ensure a rigorous and transparent literature review process. This structured approach facilitated a comprehensive overview of the current state of research in this field. Results: Through the methodology used, 64 documents were retrieved, of which 40 documents met the inclusion criteria. The reviewed papers demonstrate advancements in AI and ML for CVD detection, classification, prediction, diagnosis, and patient monitoring. Techniques such as ensemble learning, deep neural networks, and feature selection improve prediction accuracy over traditional methods. ML models predict cardiovascular events and risks, with applications in monitoring via wearable technology. The integration of AI in health care supports early detection, personalized treatment, and risk assessment, possibly improving the management of CVDs. Conclusions: The study concludes that AI and ML techniques can improve the accuracy of CVD classification, prediction, diagnosis, and monitoring. The integration of multiple data sources and noninvasive methods supports continuous monitoring and early detection. These advancements help enhance CVD management and patient outcomes, indicating the potential for AI to offer more precise and cost-effective solutions in health care.

• Monteiro, Stephanie; Oliveira, Bruno; Elvas, Luìs B. & Ferreira, Joao C. (2025). Improving healthcare data quality : a case study in a Portuguese hospital. Procedia Computer Science. 256, p. 1030–1038. doi: 10.1016/j.procs.2025.02.209. Full text in Research Archive
• Nunes, Miguel; Boné, João; Ferreira, Joao C.; Chaves, Pedro & Elvas, Luìs B. (2024). MediAlbertina: An European Portuguese medical language model. Computers in Biology and Medicine. ISSN 0010-4825. 182(November). doi: 10.1016/j.compbiomed.2024.109233. Full text in Research Archive Show summary

  Background: Patient medical information often exists in unstructured text containing abbreviations and acronyms deemed essential to conserve time and space but posing challenges for automated interpretation. Leveraging the efficacy of Transformers in natural language processing, our objective was to use the knowledge acquired by a language model and continue its pre-training to develop an European Portuguese (PT-PT) healthcare-domain language model. Methods: After carrying out a filtering process, Albertina PT-PT 900M was selected as our base language model, and we continued its pre-training using more than 2.6 million electronic medical records from Portugal's largest public hospital. MediAlbertina 900M has been created through domain adaptation on this data using masked language modelling. Results: The comparison with our baseline was made through the usage of both perplexity, which decreased from about 20 to 1.6 values, and the fine-tuning and evaluation of information extraction models such as Named Entity Recognition and Assertion Status. MediAlbertina PT-PT outperformed Albertina PT-PT in both tasks by 4–6% on recall and f1-score. Conclusions: This study contributes with the first publicly available medical language model trained with PT-PT data. It underscores the efficacy of domain adaptation and offers a contribution to the scientific community in overcoming obstacles of non-English languages. With MediAlbertina, further steps can be taken to assist physicians, in creating decision support systems or building medical timelines in order to perform profiling, by fine-tuning MediAlbertina for PT- PT medical tasks.

• Elvas, Luìs B.; Nunes, Miguel; Ferreira, Joao C. & Helgheim, Berit (2024). Hospital remote care assistance AI to reduce workload. International Journal of Computer Information Systems and Industrial Management Applications. 16(2), p. 197–209. Full text in Research Archive Show summary

  In the ever-evolving healthcare landscape, the rapid advancement of Artificial Intelligence (AI) technology presents significant opportunities for enhancing patient care and streamlining operational workflows. This study adopts the action research methodology to explore the transformative impact of AI, specifically leveraging data fusion techniques, on remote chronic patient care. Our research yields promising results, achieving 94% accuracy in predicting heart rate abnormalities using Neural Networks and 87.6% for Facial Expression classification through Convolutional Neural Networks. A critical observation from the state-of-the-art indicates that existing studies predominantly focus on remote patient monitoring, leaving a substantial gap in addressing the needs of caregivers and medical staff. Our findings demonstrate that AI, integrated into remote home care monitoring systems, has the potential to alleviate caregiver workloads by enabling continuous patient monitoring and timely alerts. The automation of data extraction and analysis enhances healthcare professionals’ decision-making capabilities, ultimately improving patient outcomes while alleviating the burdens of manual data entry.

• Phuyal, Sudip; Elvas, Luìs B.; Ferreira, Joao C. & Bista, Rabindra (2024). Harnessing Wearable Devices for Enhanced Long-Term Care: Opportunities, Challenges, and Future Directions. International Journal of Computer Information Systems and Industrial Management Applications. 16(3), p. 502–514. Full text in Research Archive Show summary

  The incorporation of wearable devices within long-term care represents a groundbreaking advancement in healthcare. These devices, leveraging sensor and intelligent technologies, possess the potential to revolutionize the supervision and handling of chronic illnesses, enriching the lives of individuals reliant on extended care. One of the primary considerations revolves around the question of whether recent technological advancements have made it feasible to deliver healthcare remotely to the elderly population. This paper conducts an exhaustive examination of wearable devices in the long-term care context, spotlighting their wide-ranging applications. It encompasses devices tailored for monitoring vital signs, mobility, medication adherence, cognitive health, and other critical facets of long-term care. Through an in-depth exploration of their capabilities and limitations, it illuminates their capacity to ease the strain on healthcare systems and enhance patient well-being. Moreover, this paper investigates the hurdles related to wearable device adoption in long-term care, addressing concerns like data security, user acceptance, and interoperability. By tackling these challenges, it endeavors to chart a path for the seamless integration of wearable technologies. Drawing insights from current research and industry trends, it identifies emerging prospects such as personalized care plans, remote monitoring, and data-driven healthcare interventions, fostering collaboration among healthcare stakeholders. By harnessing the full potential of wearable technologies, the quality of life of individuals in long-term care can be improved and also, it can contribute to achieve more efficient and sustainable healthcare systems.

• Ferreira, Joao C.; Elvas, Luìs B.; Correia, Ricardo & Mascarenhas, Miguel (2024). Enhancing EHR Interoperability and Security through Distributed Ledger Technology: A Review. Healthcare. 12(19). doi: 10.3390/healthcare12191967. Full text in Research Archive Show summary

  The management and exchange of electronic health records (EHRs) remain critical challenges in healthcare, with fragmented systems, varied standards, and security concerns hindering seamless interoperability. These challenges compromise patient care and operational efficiency. This paper proposes a novel solution to address these issues by leveraging distributed ledger technology (DLT), including blockchain, to enhance data security, integrity, and transparency in healthcare systems. The decentralized and immutable nature of DLT enables more efficient and secure information exchange across platforms, improving decision-making and coordination of care. This paper outlines a strategic implementation approach, detailing timelines, resource requirements, and stakeholder involvement while addressing crucial privacy and security concerns like encryption and access control. In addition, it explores standards and protocols necessary for achieving interoperability, offering case studies that demonstrate the framework’s effectiveness. This work contributes by introducing a DLT-based solution to the persistent issue of EHR interoperability, providing a novel pathway to secure and efficient health data exchanges. It also identifies the standards and protocols essential for integrating DLT with existing health information systems, thereby facilitating a smoother transition toward enhanced interoperability.

• Elvas, Luìs B.; Nunes, Miguel B.; Helgheim, Berit & Ferreira, Joao C. (2024). Prediction of emergency department operations with artificial intelligence : a case study. In ., IEEE (Eds.), 2024 IEEE 22nd Mediterranean Electrotechnical Conference (MELECON). IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350387025. p. 473–478. doi: 10.1109/MELECON56669.2024.10608783. Full text in Research Archive Show summary

  The emergency department (ED) of a hospital is a critical component of the healthcare system, serving as the first point of contact for patients in need of immediate medical attention. However, due to the unpredictable volume and nature of demand, managing resources in an ED can be quite challenging. The increasing demand for emergency services has resulted in overcrowding and long wait times, which can negatively impact patient outcomes and satisfaction. In this paper, we propose a solution to these problems by using artificial intelligence (AI) to identify inefficiencies in ED operations and to forecast the patient flow one week in advance. The proposed AI system resorts to machine learning algorithms to analyze the different process phases a patient has to go through while in the ED. By analyzing real-time data, the system can identify bottlenecks and recommend strategies for improving resource allocation, such as adjusting staffing levels or reassigning patients to different areas in the ED. Additionally, the system can also forecast demand for ED services, allowing hospital administrators to proactively allocate resources and ensure that patients receive timely and effective care. The proposed AI-based approach has the potential to significantly improve ED operations and patient outcomes by reducing wait times, improve resource allocation, and improve overall quality of the health service provided. By leveraging the power of AI, hospitals can provide more effective and efficient care, ultimately improving patient satisfaction and outcomes.

• Nunes, Miguel; Boné, João; Ferreira, Joao C. & Elvas, Luìs B. (2024). Health care language models and their fine-tuning for information extraction : scoping review. JMIR Medical Informatics. 12. doi: 10.2196/60164. Full text in Research Archive Show summary

  Background: In response to the intricate language, specialized terminology outside everyday life, and the frequent presence of abbreviations and acronyms inherent in health care text data, domain adaptation techniques have emerged as crucial to transformer-based models. This refinement in the knowledge of the language models (LMs) allows for a better understanding of the medical textual data, which results in an improvement in medical downstream tasks, such as information extraction (IE). We have identified a gap in the literature regarding health care LMs. Therefore, this study presents a scoping literature review investigating domain adaptation methods for transformers in health care, differentiating between English and non-English languages, focusing on Portuguese. Most specifically, we investigated the development of health care LMs, with the aim of comparing Portuguese with other more developed languages to guide the path of a non–English-language with fewer resources. Objective: This study aimed to research health care IE models, regardless of language, to understand the efficacy of transformers and what are the medical entities most commonly extracted. Methods: This scoping review was conducted using the PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) methodology on Scopus and Web of Science Core Collection databases. Only studies that mentioned the creation of health care LMs or health care IE models were included, while large language models (LLMs) were excluded. The latest were not included since we wanted to research LMs and not LLMs, which are architecturally different and have distinct purposes. Results: Our search query retrieved 137 studies, 60 of which met the inclusion criteria, and none of them were systematic literature reviews. English and Chinese are the languages with the most health care LMs developed. These languages already have disease-specific LMs, while others only have general–health care LMs. European Portuguese does not have any public health care LM and should take examples from other languages to develop, first, general-health care LMs and then, in an advanced phase, disease-specific LMs. Regarding IE models, transformers were the most commonly used method, and named entity recognition was the most popular topic, with only a few studies mentioning Assertion Status or addressing medical lexical problems. The most extracted entities were diagnosis, posology, and symptoms. Conclusions: The findings indicate that domain adaptation is beneficial, achieving better results in downstream tasks. Our analysis allowed us to understand that the use of transformers is more developed for the English and Chinese languages. European Portuguese lacks relevant studies and should draw examples from other non-English languages to develop these models and drive progress in AI. Health care professionals could benefit from highlighting medically relevant information and optimizing the reading of the textual data, or this information could be used to create patient medical timelines, allowing for profiling.

• Elvas, Luìs B.; Gomes, Sara; Ferreira, Joao C.; Rosário, Luís Brás & Brandão, Tomás (2024). Deep learning for automatic calcium detection in echocardiography. BioData Mining. 17(1). doi: 10.1186/s13040-024-00381-1. Full text in Research Archive Show summary

  Cardiovascular diseases are the main cause of death in the world and cardiovascular imaging techniques are the mainstay of noninvasive diagnosis. Aortic stenosis is a lethal cardiac disease preceded by aortic valve calcification for several years. Data-driven tools developed with Deep Learning (DL) algorithms can process and categorize medical images data, providing fast diagnoses with considered reliability, to improve healthcare effectiveness. A systematic review of DL applications on medical images for pathologic calcium detection concluded that there are established techniques in this field, using primarily CT scans, at the expense of radiation exposure. Echocardiography is an unexplored alternative to detect calcium, but still needs technological developments. In this article, a fully automated method based on Convolutional Neural Networks (CNNs) was developed to detect Aortic Calcification in Echocardiography images, consisting of two essential processes: (1) an object detector to locate aortic valve – achieving 95% of precision and 100% of recall; and (2) a classifier to identify calcium structures in the valve – which achieved 92% of precision and 100% of recall. The outcome of this work is the possibility of automation of the detection with Echocardiography of Aortic Valve Calcification, a lethal and prevalent disease.

• Elvas, Luìs B.; Nunes, Miguel; Francisco, Bruno; Gonçalves, Frederica; Martins, Ana Lucia & Ferreira, Joao C. (2023). Points of interest in smart cities and visitor behavior. In Martins, Ana Lucia; Ferreira, Joao C.; Kocian, Alexander; Tokkozhina, Ulpan; Helgheim, Berit & Bråthen, Svein (Ed.), Intelligent Transport Systems : 7th EAI International Conference, INTSYS 2023, Molde, Norway, September 6-7, 2023, Proceedings. Springer. ISSN 9783031493799. p. 69–91. doi: 10.1007/978-3-031-49379-9_4. Full text in Research Archive Show summary

  Home Intelligent Transport Systems Conference paper Points of Interest in Smart Cities and Visitor Behavior Luís B. Elvas, Miguel Nunes, Bruno Francisco, Frederica Gonçalves, Ana Lucia Martins & Joao Carlos Ferreira Conference paper First Online: 12 December 2023 15 Accesses Part of the Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering book series (LNICST,volume 540) Abstract Smart cities leverage technology and data to enhance the quality of urban life, including the management of points of interest (POIs) and visitor experiences. This paper explores the relationship between POIs and visitor behavior in smart cities, examining the impact of technology-driven solutions on understanding, analyzing, and optimizing visitor experiences. It highlights the importance of data-driven approaches in identifying and managing POIs, enhancing visitor satisfaction, and driving economic growth. The paper reviews existing literature, discusses key concepts, and presents case studies to illustrate the role of POIs in smart cities and their influence on visitor behavior. Our major contribution is a data driven approach to extract useful information from real data to municipality decisions and understand the problem. It concludes with recommendations for future research and practical implications for city planners, policymakers, and tourism authorities.

• Elvas, Luìs B.; Nunes, Miguel; Ferreira, Joao C.; Francisco, Bruno & Afonso, Jose A. (2023). Georeferenced Analysis of Urban Nightlife and Noise Based on Mobile Phone Data. Applied Sciences. 14(2024)(1). doi: 10.3390/app14010362. Full text in Research Archive Show summary

  Urban environments are characterized by a complex soundscape that varies across different periods and geographical zones. This paper presents a novel approach for analyzing nocturnal urban noise patterns and identifying distinct zones using mobile phone data. Traditional noise-monitoring methods often require specialized equipment and are limited in scope. Our methodology involves gathering audio recordings from city sensors and localization data from mobile phones placed in urban areas over extended periods with a focus on nighttime, when noise profiles shift significantly. By leveraging machine learning techniques, the developed system processes the audio data to extract noise features indicative of different sound sources and intensities. These features are correlated with geographic location data to create comprehensive city noise maps during nighttime hours. Furthermore, this work employs clustering algorithms to identify distinct noise zones within the urban landscape, characterized by their unique noise signatures, reflecting the mix of anthropogenic and environmental noise sources. Our results demonstrate the effectiveness of using mobile phone data for nocturnal noise analysis and zone identification. The derived noise maps and zones identification provide insights into noise pollution patterns and offer valuable information for policymakers, urban planners, and public health officials to make informed decisions about noise mitigation efforts and urban development.

• Elvas, Luìs B.; Nunes, Miguel; Afonso, José Augusto; Helgheim, Berit & Francisco, Bruno (2023). Mining tourists’ movement patterns in a city. In Martins, Ana Lucia; Ferreira, Joao C.; Kocian, Alexander; Tokkozhina, Ulpan; Helgheim, Berit & Bråthen, Svein (Ed.), Intelligent Transport Systems : 7th EAI International Conference, INTSYS 2023, Molde, Norway, September 6-7, 2023, Proceedings. Springer. ISSN 9783031493799. p. 105–125. doi: 10.1007/978-3-031-49379-9_6. Full text in Research Archive Show summary

  Although tourists generate a large amount of data (known as “big data”) when they visit cities, little is known about their spatial behavior. One of the most significant issues that has recently gained attention is mobile phone usage and user behavior tracking. A spatial and temporal data visualization approach was established with the purpose of finding tourists’ footprints. This work provides a platform for combining multiple data sources into one and transforming information into knowledge. Using Python, we created a method to build visualization dashboards aiming to provide insights about tourists’ movements and concentrations in a city using information from mobile operators. This approach can be replicated to other smart cities with data available. Weather and major events, for instance, have an impact on the movements of tourists. The outputs from this work provide useful information for tourism professionals to understand tourists’ preferences and improve the visitors’ experience. Management authorities may also use these outputs to increase security based on tourists’ concentration and movements. A case study in Lisbon with 4 months data is presented, but the proposed approach can also be used in other cities based on data availability. Results from this case study demonstrate how tourists tend to gather around a set of parishes during a specific time of the day during the months under study, as well as how unusual circumstances, namely international events, impact their overall spatial behavior.

• Elvas, Luìs B.; Nunes, Miguel; Francisco, Bruno & Domingues, Nuno (2023). City mobility and night life monitor. In Martins, Ana Lucia; Ferreira, Joao C.; Kocian, Alexander; Tokkozhina, Ulpan; Helgheim, Berit & Bråthen, Svein (Ed.), Intelligent Transport Systems : 7th EAI International Conference, INTSYS 2023, Molde, Norway, September 6-7, 2023, Proceedings. Springer. ISSN 9783031493799. p. 126–150. doi: 10.1007/978-3-031-49379-9_7. Full text in Research Archive Show summary

  Home Intelligent Transport Systems Conference paper City Mobility and Night Life Monitor Luís B. Elvas, Miguel Nunes, Bruno Francisco & Nuno Domingues Conference paper First Online: 12 December 2023 15 Accesses Part of the Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering book series (LNICST,volume 540) Abstract This paper presents an Internet of Things (IoT) system designed to collect and analyse information regarding the travel patterns and movements of individuals in densely populated locations, in the context of smart cities. People’s movements are retrieved from coarse-grained aggregated cellular network data without collecting sensitive information from mobile devices and users. These data were provided by a Portuguese cellular operator to the Lisbon City Council to characterize people movements in the city. In this sense, the mobile phones act as useful sensor devices for collecting rich spatiotemporal information about human movement patterns. The purpose of this research work is to create a machine learning-based data-driven approach that is able to receive anonymised data from telecommunication operators to provide a big picture about citizen mobility in the city and to identify patterns based on the collected data, in order to provide relevant information for city planning and events coordination. Some of the main applications of the proposed system are the coordination of big events and the management and control of commuting traffic.

• Ferreira, Joao C.; Francisco, Bruno; Elvas, Luìs B.; Nunes, Miguel & Afonso, Jose A. (2023). Predicting People’s Concentration and Movements in a Smart City. Electronics (Basel). 13(2024)(1). doi: 10.3390/electronics13010096. Full text in Research Archive Show summary

  first_pagesettingsOrder Article Reprints Open AccessArticle Predicting People’s Concentration and Movements in a Smart City by Joao C. Ferreira 1,2,3ORCID,Bruno Francisco 1,Luis Elvas 1,2,3,Miguel Nunes 1 andJose A. Afonso 4,*ORCID 1 Instituto Universitário de Lisboa (ISCTE-IUL), ISTAR, 1649-026 Lisboa, Portugal 2 Inov Inesc Inovação—Instituto de Novas Tecnologias, 1000-029 Lisbon, Portugal 3 Department of Logistics, Molde University College, 6410 Molde, Norway 4 CMEMS-UMinho/LABBELS—Associate Laboratory, University of Minho, 4800-058 Guimarães, Portugal * Author to whom correspondence should be addressed. Electronics 2024, 13(1), 96; https://doi.org/10.3390/electronics13010096 Submission received: 25 October 2023 / Revised: 18 December 2023 / Accepted: 21 December 2023 / Published: 25 December 2023 (This article belongs to the Special Issue AI Technologies and Smart City) Downloadkeyboard_arrow_down Browse Figures Versions Notes Abstract With the rapid growth of urbanization and the proliferation of mobile phone usage, smart city initiatives have gained momentum in leveraging data-driven insights to enhance urban planning and resource allocation. This paper proposes a novel approach for predicting people’s concentration and movements within a smart city environment using mobile phone data provided by telecommunication operators. By harnessing the vast amount of anonymized and aggregated mobile phone data, we present a predictive framework that offers valuable insights into urban dynamics. The methodology involves collecting and processing location-based data obtained from telecommunication operators. Using machine learning techniques, including clustering and spatiotemporal analysis, we developed models to identify patterns in people’s movements and concentration across various city regions. Our proposed approach considers factors such as time of day, day of the week, and special events to capture the intricate dynamics of urban activities. The predictive models presented in this paper demonstrate the ability to predict areas of high concentration of people, such as commercial districts during peak hours, as well as the people flow during the time. These insights have significant implications for urban planning, traffic management, and resource allocation. Our approach respects user privacy by working with aggregated and anonymized data, ensuring compliance with privacy regulations and ethical considerations. The proposed models were evaluated using real-world mobile phone data collected from a smart city environment in Lisbon, Portugal. The experimental results demonstrate the accuracy and effectiveness of our approach in predicting people’s movements and concentration. This paper contributes to the growing field of smart city research by providing a data-driven solution for enhancing urban planning and resource allocation strategies. As cities continue to evolve, leveraging mobile phone data from telecommunication operators can lead to more efficient and sustainable urban environments.

View all works in NVA

• Ferreira, Joao C.; Elvas, Luìs B.; Correia, Ricardo & Mascarenhas, Miguel (2025). Tokenization of electronic health records and healthcare data: enhancing security and privacy while enabling usability. Health and Technology. ISSN 2190-7188. 15(6), p. 1011–1020. doi: 10.1007/s12553-025-01012-3. Full text in Research Archive Show summary

  Abstract Purpose The purpose of this paper is to examine the role of tokenization in protecting Electronic Health Records (EHRs) and healthcare data, with a focus on ensuring data security and privacy compliance. Tokenization, which replaces sensitive patient data with unique identifiers (tokens), is explored as a means to prevent unauthorized access while maintaining data utility for healthcare operations, research, and analytics. Additionally, the paper aims to assess how tokenization helps healthcare organizations comply with regulations like HIPAA, and how it facilitates secure data sharing and analytics. Methods This state-of-the-art review analyzes the latest advancements in tokenization techniques, particularly their applications in safeguarding patient data during storage, transmission, and analysis. The review includes an exploration of tokenization’s role in secure data exchange, system interoperability, and the integration of natural language processing (NLP) for analyzing de-identified healthcare data. Various methodologies and real-world implementations of tokenization in healthcare are comprehensively assessed. Results The analysis reveals that tokenization effectively enhances data security and privacy, supporting compliance with stringent regulatory frameworks like HIPAA. Furthermore, tokenization enables the secure exchange of healthcare data across systems while maintaining interoperability. The integration of NLP for de-identified data analysis further expands its application in research and analytics without compromising patient confidentiality. Conclusions The review highlights the potential of tokenization to balance data security and usability within healthcare systems. It improves patient care by facilitating secure data sharing and supporting advanced analytics, while simultaneously ensuring operational efficiency and regulatory compliance. Tokenization emerges as a crucial tool in enhancing the protection and utility of healthcare data.

• Elvas, Luìs B.; Ferreira, Joao C. & Helgheim, Berit (2025). Hospital remote care assistance AI to reduce workload. In Sakalauskas, Virgilijus; Bajaj, Anu; Abraham, Ajith; Madhavi, K. Reddy & Mishra, Pooja Manghirmalani (Ed.), Bio-Inspired Computing : Proceedings of the 14th International Conference on Innovations in Bio-Inspired Computing and Applications (IBICA 2023) held during December 14-15, 2023, Volume 4: Machine Learning Applications. Springer. ISSN 9783031789465. p. 61–72. doi: 10.1007/978-3-031-78946-5_6. Full text in Research Archive Show summary

  This research aims to create a remote system to reduce the Hospital load for patients with chronic diseases at Home or Recover Centers. A framework of shared Artificial intelligence services is proposed to provide a secure environment for remote patient monitoring. A literature review using the PRISMA methodology provides a knowledge systematization. A new data fusion process with data, image, text, and voice allows for improving remote patient data collected. We apply Design Science Research with CRISP-DM approach to handling the data process. We are developing specific modules to fuse and handle image, voice, text and data. Our model is being validated and applied in a real case from a Portuguese hospital using Alzheimer’s patients. In our research work, we propose a framework for secure information sharing among patients’ data and improving health care services. This approach allows the creation of big data systems in a fusion of image, text, voice, and data. The quick development of AI and associated technologies would assist healthcare providers in enhancing patient value and streamlining operational procedures. In this study, we examined how AI technology with data fusion for chronic patients allows professional hospital visits and reduces their workload. Effective management of these opportunities and challenges requires the joint knowledge and tenacity of all healthcare industry players.

• Elvas, Luìs B. (2025). Structuring the unstructured : an AI-driven approach to electronic health record transformation for clinical decision support. Høgskolen i Molde - Vitenskapelig høgskole i logistikk. ISSN 9788279623908. Full text in Research Archive
• Elvas, Luìs B. & Mawrides, Emmanuel Kafti (2024). Empowering patient-centric health data exchange : a blockchain-based framework for secure and efficient data sharing. Høgskolen i Molde - Vitenskapelig høgskole i logistikk. ISSN 9788279623717. Full text in Research Archive Show summary

  In the healthcare sector, the sharing of sensitive health data poses substantial privacy and security challenges that must be navigated with utmost care. With health data recognized by the European Commission as a unique and valuable resource for both retrospective and prospective research, and the Organisation for Economic Co-operation and Development (OECD) advocating for robust health data governance, there is a clear directive to harmonize individual privacy with the need for data accessibility. This paper proposes a conceptual blockchain framework designed to empower patients in the management and sharing of their medical information. Through this framework, patients would assume the role of data custodians, with the authority to grant access to their health records as they see fit. By leveraging blockchain technology's inherent qualities of decentralization, immutability, and transparency, our framework ensures that data sharing is secure, patient-centered, and conducted under the patient's consent. This approach not only upholds the privacy and autonomy of individuals but also streamlines the data sharing process. It is particularly advantageous for research centers, which can access necessary data without the procedural delays associated with traditional ethical approval processes. Our conceptual framework aims to redefine health data exchange, prioritizing patient control and trust, while fostering a more efficient and collaborative research environment.

View all works in NVA