ESS - BBB - Comunicações em eventos científicos
Permanent URI for this collection
Browse
Recent Submissions
- From controlled to chaotic: Disparities in laboratory vs real-world stress detectionPublication . Ferreira, Simão; Rodrigues, Fátima; Kallio, Johanna; Coelho, Filipe; Kyllonen, Vesa; Rocha, Nuno; Rodrigues, Matilde A.; Vildjiounaite, Elena; Ferreira, Simão; Rodrigues, MatildeThis paper explores the discrepancies between laboratory and real-world stress detection, emphasizing the pronounced differences in data loss, data preprocessing, feature design, and classifier selection. Laboratory studies offer a controlled environment that optimizes data quality, whereas real-world settings introduce chaotic and unpredictable elements, coupled with a diverse range of human behaviours, resulting in substantial data loss and compromised data quality. We discuss the development of stress detectors for two distinct types of data: physiological and behavioural. We also address the specific challenges associated with designing effective stress detection systems for each data type and compare the features and classifiers used in both laboratory and real-world contexts. Additionally, this paper proposes future research directions aimed at crafting stress detectors that are robust and effective in real-life scenarios.
- Virtual journey through an immersive interactive environment: A sensory exploration of an artistic space modeled by dynamic painting and emotional music by Domingos MateusPublication . Gomes, Paulo Veloso; Sá, Vítor J.; Donga, João; Marques, António; Mateus, Domingos; Machado Veloso Gomes, Paulo Sérgio; Rucha das Dores da Costa Donga, João Paulo; Pereira da Silva Marques, António JoséThis study explores the sensory impact of a virtual journey thr ough an immersive, interactive envir onment inspir ed by the artistic work of Domingos Mateus. It addr esses how sensory experiences in virtual spaces, using dynamic visual and auditory stimuli, can enhance emotional engagement and spatial awareness. This study aims to investigate the ef fectiveness of combining dynamic painting and emotionally resonant music within an interactive, digital envir onment, aiming to evoke a deeper connection between viewers and artistic content. A multi-sensory museum was implemented, blending interactive visual components (dynamic painting) with a custom musical scor e designed to elicit emotional responses. Participants navigated the virtual envir onment using VR equipment, cr eating a fully immersive experience. Sensory responses wer e monitored using real-time biofeedback to gauge emotional engagement and pr esence. Findings suggest combining interactive visual art and music significantly enhances users' emotional engagement and immersive experience. Participants reported heightened spatial pr esence, with biofeedback data indicating incr eased emotional ar ousal during key moments. This sensory appr oach is potentially used in therapeutic and educational settings, wher e emotional and sensory stimulation ar e beneficial. The study underscores the power of immersive envir onments in enhancing the user’ s connection to artistic expr essions and fostering memorable experiences.
- Optimization of surgical scheduling: Predicting surgery time duration using machine learningPublication . Malheiro, Soraia; Faria, Brígida; Dias, Celeste; Faria, Brigida MonicaThe operating room (OR) is a highly specialized hospital department that requires a large amount of resources which has a high impact on hospital funding (1). The OR is an essential area for the hospital operation and its management must guarantee the best efficiency and the highest quality of patient care. Despite some initiatives already implemented to meet the demand for surgical treatment, as described in the European Commission’s 2021 report (2), waiting times for surgery in Portugal have increased in the last Currently, this prediction is essentially based on the surgeon’s experience in a particular surgical procedure and may not take other variables into account (1). The aim of this study is to predict more accurately the duration of the surgeries in the specialties of General Surgery, Orthopedics and Urology by developing a model based in machine learning techniques with data from clinical records of surgical cases. Methods: The sample of this study includes data from surgical cases performed in a hospital center. The following cases were excluded: Surgeries with patients under the age of 18; without a defined preoperative diagnosis; unspecified surgical specialties; no record of the start and/or end time of surgery and surgeries that took place on an outpatient basis. Multiple Linear Regression (MLR) and Random Forest (RF) techniques were applied to develop the model. Accuracy in predicting the duration of surgeries can optimize the OR occupancy and at the same time decrease the waiting time experienced by the patients. decade. Surgical scheduling is fundamental in the OR management (3). One of the challenges related to surgical scheduling is the prediction of surgery duration, which is essential for the allocating OR occupancy times.
- Modelling therapeutic response in asthmatic adults: a previous exploratory analysisPublication . Alves, Cristina; Faria, Brígida Mónica; Alves, Sandra Maria; Ferreira, Jorge; Faria, Brigida Monica; Alves, Sandra MariaAsthma is a respiratory disease characterized by chronic inflammation of the airways. Effective asthma management is essentially based on choosing the appropriate treatment for each individual (1). Data science and machine learning models offer valuable insights and enhance the outcomes achieved in asthma management (2). The main objective is to develop predictive models for therapy response in patients with asthma, and secondarily to identify clinical, functional and biological characteristics that influence this response. Data from fifty adults with asthma were analyzed, collecting information on anthropometric, clinical, functional, biological, therapeutic, occupational, and allergen exposure factors. The study followed the “Knowledge Discovery in Databases, KDD” methodology. The sample consisted of 50 asthmatic adult participants, aged between 21 and 81 years old mean age=54.02 (s=14.5), from which 20 (40%) were male and 30 (60%) were female. The analysis of the characteristic symptoms of asthma (dyspnea, cough, wheezing and chest tightness), reveals a statistically significant improvement (p<0.001) of all these symptoms after the treatment. The asthma control test, the life quality questionnaire and the asthma and allergic rhinitis control test evaluated before and after treatments, demonstrate a statistically significant difference (p=0.023, p <0.001 and p<0.001, respectively). On respiratory function, only FVC reveals a significant difference (p=0.409), after treatment. However, the average did not reach the minimal important difference (MID) of 200ml. The average number of exacerbations and SU recurrences difference was also significant in both cases (p<0.01), reaching MID (>50%). The majority of the individuals in this group had a positive, clinically important response to treatment. This result may be because they have severe atopic asthma, and Th2-High endotype, and for that reason they are undergoing more differentiated treatments, such as biological treatments.
- Contributions to the cross-cultural validation of “A survey of pharmacist knowledge, attitudes, utilization and barriers toward artificial intelligence”: translation and back translationPublication . Gerardo, Sofia; Pimenta, Rui; Alves, Sandra Maria; Alves, Sandra MariaThe use of Artificial Intelligence (AI) is rapidly transforming various fields, and pharmacy is no exception. AI is increasingly being used to automate, optimize, and personalize various tasks in pharmacy practice, from drug discovery to dispensing to patients. In Community Pharmacy, in addition to these possibilities, it allows for personalized and focused patient care through the selection of more appropriate and personalized therapies, with a lower probability of prescription errors and drug interactions, as well as monitoring of therapy adherence. Despite these potential benefits, its implementation in the pharmaceutical field, as well as in other areas of healthcare, should be carefully considered, as ethical and regulatory issues may pose obstacles. Likewise, the perspective and experience of each professional, which remain highly personal, especially in patient care, should not be overlooked. Therefore, it is increasingly important to know the knowledge, attitudes, utilization, and barriers concerning AI. Firstly, knowledge, as this concept encompasses the level of awareness and understanding that individuals or organizations have regarding AI technologies. Attitudes, which refers to the perceptions, feelings, and predispositions towards AI. It includes both positive and negative sentiments, such as excitement about AI's potential benefits, concerns about ethical implications, and fears about job displacement. Finally, the barriers or obstacles that hinder the adoption and effective utilization of AI. Barriers can be technical, such as lack of expertise or inadequate infrastructure; financial, such as high costs of implementation; or cultural, such as resistance to change or lack of trust in AI systems. This study aimed to translate and validate a survey instrument designed to explore the attitudes of Community Pharmacy professionals towards the implementation of Artificial Intelligence (AI) in their field. According to the COSMIN methodology, the initial translation of the construct from its original language, English, into European Portuguese was executed by two independent translators possessing comprehensive understanding of the questionnaire concepts. Both translators are bilingual healthcare professionals, who perform functions in a hospital environment and with European Portuguese being their native language. In the subsequent step, the two acquired versions were juxtaposed, leading to the formation of a consensus version endorsed by specialists, considering the new context where the construct will be applied, without excluding the original version. In cases where there are discrepancies between the two translations, the expert panel discusses the alternatives and decides on the most suitable option. The questionnaire translation process culminates with back-translation, wherein the consensus version obtained is rendered back into the original language, English, by a bilingual translator. The resultant back-translation should closely mirrors the original questionnaire, signifying the efficacy of the content translation process. Additionally, reliability testing methods like test-retest reliability and internal consistency checks help verify the stability and consistency of the survey results. To assess consensus among different questionnaire versions, techniques such as inter-rater reliability, the Delphi method, and agreement indices are essential. These processes ensure that the survey items are interpreted consistently across different respondents and that any subjective judgments are reliably measured. The results of the translated version questionnaire maintain the intended constructs and adequately capture attitudes toward AI implementation among Community Pharmacy professionals. Understanding the attitudes and perceptions of pharmacy professionals towards AI implementation is crucial for informing policy decisions, designing targeted interventions, and facilitating the successful integration of AI technologies into pharmacy practice. It is intended that this questionnaire contributes to the growing body of literature on AI in healthcare and serves as a foundation for further investigations into this evolving field. Future work includes the validation of the PT-EU questionnaire.
- Randomization in clinical trials: a biostatistician taskPublication . Coelho, Heitor; Albuquerque, João; Alves, Sandra Maria; Alves, Sandra MariaRandomized clinical trials stand as the golden standard for evidence-based clinical investigation on humans, providing robust methodologies to evaluate new interventions (1). By employing stochastic processes to allocate participants to treatment and control groups, randomization minimizes bias and enhances the validity of results (1,2). This ensures unpredictability in treatment assignment, mitigating selection, response, and confounding biases, thus safeguarding the statistical integrity of the study (1,3,4). Various randomization techniques address specific trial needs (5). Simple randomization is the most straightforward, offering equal allocation probabilities but risking imbalance in small samples (6). Block randomization ensures numerical balance across groups but may fail to address covariate comparability (7). Stratified randomization improves group balance concerning key covariates but demands careful variable selection to prevent empty strata (5,6). Adaptive randomization adjusts allocation probabilities during the trial to manage imbalances but introduces complexity and potential predictability (1,8). Selecting the appropriate randomization method requires careful consideration of study design, sample size, and confounding factors to ensure reliable results (5). Beyond method selection, effective implementation is crucial, necessitating adherence to best practices outlined in the CONSORT 2010 guidelines (1). Proper documentation of allocation sequence generation, masking procedures, and protocol monitoring ensures reproducibility and integrity (1). This session will introduce randomization in clinical trials, present different methodologies, and explore their strengths and weaknesses.
- Predicting treatment response in wet age-relatedd macular degeneration through OCT biomarkersPublication . Sousa, Vânia; Carneiro, Ângela; Faria, Brígida Mónica; Faria, Brigida MonicaWet age-related Macular Degeneration (AMD), characterized by macular neovascularization that leads to fluid leakage and retinal hemorrhage causing severe and sometimes irreversible visual damage, is one of the biggest causes of blindness in developed countries (1). Treatment with anti-vascular endothelial growth factors (anti- VEGF) has been revolutionary, but not all patients respond completely to treatment and have unmet clinical needs (2). The schedule and response to these treatments is a burden for patients/carers and hospitals (2). The use of optical coherence tomography (OCT) has become a valuable tool for diagnosing and monitoring this pathology. There are biomarkers in the retina detectable with state-of-the-art OCT that may be associated with visual recovery after treatment with anti-VEGF (3). Objectives: Identify the biomarkers with the most influence on the response to treatment and develop various supervised learning algorithms to predict the response to treatment in order to better adapt the treatment to each patient, reducing the burden that the tight schedule of these injections has on patients/caregivers, hospitals and health professionals. We collected general data, visual acuity and OCT from patients with wet AMD undergoing treatment with anti-VEGF injections, followed at the São João Local Health Unit for 3 years. A statistical analysis and study of the variables with the greatest weight in the response to treatment will be carried out. With these variables, we intend to use various supervised learning algorithms to see if it is possible to create a model with a good accuracy rate for predicting the response to treatment. We have collected data from 98 eyes of 81 patients, 29 female (35.8%) and 52 male (64.2%) with mean age of 76.93 ± 7.6 years, with mean initial visual acuity of 60.15 letters and 58.76% of eyes with type I membrane. With this data we will identify the biomarkers with the most influence on the response to treatment and select the algorithm with the best model evaluation metrics. By identifying biomarkers and selecting an algorithm, we can find ways to improve patient treatment. Making this study multicentric would be an improvement, but data collection always requires specialized professionals and is time-consuming.
- Exploring coping profiles in informal caregivers of people with dementia through an online eHealth platformPublication . Gouveia, Tânia; Viana, João; Faria, Brígida; Teles, Soraia; Faria, Brigida MonicaInformal caregivers (IC) of people with dementia (PwD) handle multiple stressors with an acknowledged impact on their physical and mental health. health (1,2). Coping strategies may be viewed as the techniques employed by ICs to minimize distress associated with the caregiving experience. iSupport emerges as an online self-guided program developed by the World Health Organization (3) to provide support and training for ICs of PwD to minimize the negative impact of caring for a PwD. This program has been translated and culturally adapted for the Portuguese population and tested for its usability and feasibility (4,5). iSupport-Portugal1 is, being explored for its potential as an intervention-research tool, offering support to ICs of PwD but also serving as a remote measurement tool to collect data on dementia care dyads. Research goal: To describe coping styles and facets among IC of PwD and to identify profiles of IC in relation to coping styles based on underlying data pattens from demographic, clinical and psychosocial variables pertaining to both the IC and PwD. This study follows an observational cross-sectional design. Participants are IC registered on iSupport- Portugal between February 2023 and February 2024, who meet eligibility criteria. Data on the IC and PwD are collected upon registration via user’s accounts. Data analysis: Unsupervised learning methods, such as cluster analysis, enable the identification of latent patterns or underlying structures in dimensional data. Their ability to discover similarities and differences in information makes them an ideal solution for exploring IC profiles in relation to coping strategies. Expected results and Implications: Coping is a modifiable dimension with known implications to well-being outcomes for both the IC and the PwD. Identifying IC profiles in relation to coping strategies is fundamental to design more targeted interventions addressing the specific needs of the IC-PwD dyads.
- Swallowing assessment in a clinical context: artificial intelligence applicationsPublication . Torres, Teresa; Faria, Brígida; Araújo, Pedro; Oliveira, Alexandra; Alves, Elisa; Faria, Brigida Monica; Oliveira, AlexandraEating and swallowing are intricate actions comprehending both voluntary and reflexive movements, engaging over thirty nerves and muscles (1). Dysphagia occurs when the normal swallowing process is compromised, increasing the risk of the swallowed material entering the larynx, and causing complications for the patient (2). Cervical auscultation (CA) is a clinical method used to evaluate the pharyngeal phase of swallowing by listening to the sounds of swallowing-related respiration. However, the reliability of CA is susceptible to the subjectivity and experience of the speech therapist (3). The application of artificial intelligence (AI) tools in healthcare has the potential to support healthcare workers with a variety of tasks. It can be used for disease prediction and diagnostics treatment, outcome prediction and prognosis evaluation (4). Contribute to the development of an AI tool to aid speech therapists in their daily evaluation of deglutition in patients, applying machine learning algorithms to process and analyse the sound recorded during cervical auscultation. Using an electronic stethoscope, audio samples were recorded from individuals with and without pathology while they swallowed different liquid quantities (5 mL and 10 mL) and consistency (moderately thick and solid). The audio is divided into three sections (5) and pre- processed to remove unnecessary noise. Data is classified using machine learning algorithms and the models will be evaluated according to precision, accuracy, sensibility, confusion matrix and ROC curve. Previous work in this field reported by Santoso et al. (6) using supervised machine learning algorithms obtained promising results in swallowing detection. In this study, our dataset is composed of 87 samples of patients, and similar results are to be expected. AI can offer an objective and quantitative analysis of swallowing sounds, potentially providing more accurate and consistent results compared to subjective assessments. Additionally, it can identify complex patterns in swallowing sounds that may be challenging for less experienced speech therapists to recognize.
- Comparing time series forecasting models for health indicators: A clustering analysis approachPublication . Vinhal, Cláudia; Oliveira, Alexandra; Faria, Brígida; Nascimento, Ana Paula; Pimenta, Rui; Oliveira, Alexandra; Faria, Brigida Monica; Pimenta, RuiTime series are the sequence of observations ordered by equal time intervals, crucial for understanding causality, trends, and forecasts. Its analysis can be applied to several areas, such as engineering, finance, and health (1,2). One problem with the time series study is clustering, mainly understanding when two parametric time series are considered similar (3). The sum of mortality and morbidity, referred to as “Burden of Disease”, is measured by a metric called “Disability Adjusted Life Years” (DALYs) (4). These indicators are direct measures of health care needs, reflecting the global burden of disease in the population, and are crucial for public health study and surveillance (5). DALYs can be represented by Autoregressive Integrated Moving Averages (ARIMA) models, and in this context understanding clusters is crucial. The primary goal is to compare different distance measures between ARIMA processes when used in clustering techniques. The study begins by exploring the temporal characteristics of DALYs, highlighting underlying patterns and trends. Then, ARIMA models are applied to represent and describe the time series. It’s on this representation of the time series that the Piccolo, the Maharaj, and the LPC distance measures are applied to use clustering techniques and identify clusters. Additionally, 8 distinct cluster validation metrics are used. Specific to 48 European countries, the results show that the choice of distance measure can greatly influence clustering outcomes and the number of clusters formed. While certain methods revealed geographic patterns, other factors, such as cultural or economic similarities, also influence cluster formation. These insights contribute to advancing the field of public health surveillance and intervention, ultimately aiming to alleviate the global burden of disease. This study offers insights into applying ARIMA processes in clustering techniques for analysing temporal health data. By comparing different distance measures, this research improves our understanding of underlying patterns and trends in health indicators over time.