Browsing by Issue Date, starting with "2023-12-04"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
- 2ARTs – Decision Support System for Exercise and Diet Prescriptions in Cardiac Recovery PatientsPublication . Pereira, Andreia Alexandra Sousa; Martinho, Ricardo Filipe Gonçalves; Rijo, Rui Pedro Charters Lopes; Grilo, Carlos Fernando de AlmeidaThe global health care system is faced with a variety of complicated challenges, ranging from limited access and increasing expenses to an aging population causing increased pressure on healthcare systems. Healthcare professionals are seeking alternative approaches to provide fair access and sustain high-quality care for everyone as a result of these challenges. Patients have historically been restricted from accessing essential healthcare services due to traditional barriers like geographic distance, financial and resource limitations. Innovative solutions to these problems are starting to take shape, thanks to the growth of eHealth platforms that use technology to improve patient care. Through a comprehensive study of existing solutions in the healthcare domain, particularly in cardiology, we identified the need for a Decision Support System (DSS) that would empower physicians with valuable insights and facilitate informed physical and diet prescribing practices into Cardiac Rehabilitation Programmes (CRPs). The major goal of 2ARTs’ project is to create and implement a cardiac rehabilitation platform into a hospital's infrastructure. A key aspect of this platform is the integration of a decision support system designed to provide physicians with valuable information when prescribing individualized treatment prescriptions for each patient, minimizing the potential of human error. The DSS uses algorithms and predictive models to classify patients into distinct groups based on their features and medical history. This classification provides critical insights and additional knowledge to doctors, allowing them to make informed judgments regarding the most effective treatment options for each patient's cardiac rehabilitation journey. By using the power of data-driven analytics and machine learning, the DSS enables doctors to better understand each patient's needs and personalize treatment actions accordingly. In order to achieve the best possible results aligned with the goals of the project, a variety of approaches based on comprehensive studies were explored, specifically feature selection and feature reduction methods, where their performance metrics were evaluated, seeking the most effective solution. It was through this thorough analysis that Principal Component Analysis (PCA) emerged as the standout choice. PCA not only demonstrated superior outcomes in evaluation metrics, but also showcased excellent compatibility with the selected clustering algorithm along with the best results after an expert analysis. Moreover, with the analysis of the data types and features the dataset had, the K-Means algorithm produced the best results and was more adaptable to our dataset. We were able to identify useful insights and patterns within the data by employing both PCA and K-Means, opening the way for more accurate and informed decision-making in the 2ARTs project.
- Design for Oceanic Health & Wellbeing Through Sustainable Practices. An Online Platform for Educational Content Creation and Incentivising Innovation, Driven by Open Science and Blockchain TechnologyPublication . Flynn, Joseph Frederick; Pernencar, Cláudia Alexandra da Cunha; Mouga, Teresa Margarida Lopes da SilvaThe health of our oceans is under threat from climate change, ecosystem degradation, and pollution, posing significant risks to global human health and well-being. The COVID-19 pandemic has highlighted the consequences of environmental mismanagement, underscoring the urgency of sustainable practices. This study explores the potential for building environmentally sustainable and regenerative foundations for society, fostering green growth. New ocean farming technologies and algae research hold immense promise across various domains, including health, biomaterials, medicine, ecosystem services, food production, circular economies, global industries, and design applications to name a few. By sequestering carbon dioxide from the atmosphere, the oceans mitigate climate change, supply food, and support vital economic sectors like tourism and fishing, while nurturing diverse ecosystems. In the pursuit of environmental sustainability, the world needs eco-friendly and sustainable industries like regenerative ocean farming and land-based sustainable agriculture. Additionally, the oceans’ economic value, stemming from their rich biodiversity, underscores their importance. This thesis addresses a critical need for an accessible online platform that consolidates information on ocean-related innovations. The platform aims to inspire individuals to develop solutions, fostering innovation in environmental recovery, healthcare, design, food production, industry, and the ocean economy. By providing an Open Educational Resource (OER) integrated with crowdfunding capabilities, users can learn, share, fund, and innovate, thereby advancing sustainable economic and environmental solutions. Blockchain technology and cryptocurrencies offer the potential for decentralized governance, increasing user engagement and trust within the platform. The envisioned platform seeks to empower individuals to drive sustainable change and contribute to a healthier planet.
- CASE ID DETECTION IN UNLABEL LED EVENT LOGS FOR PROCESS MININGPublication . Vicente, André Alexandre dos Santos; Rijo, Rui Pedro Charters Lopes; Martinho, Ricardo Filipe Gonçalves; Grilo, Carlos Fernando de AlmeidaIn the realm of data science, event logs serve as valuable sources of information, capturing sequences of events or activities in various processes. However, when dealing with unlabelled event logs, the absence of a designated Case ID column poses a critical challenge, hindering the understanding of relationships and dependencies among events within a case or process. Motivated by the increasing adoption of data-driven decision-making and the need for efficient data analysis techniques, this master’s project presents the "Case ID Column Identification Library" project. This library aims to streamline data preprocessing and enhance the efficiency of subsequent data analysis tasks by automatically identifying the Case ID column in unlabelled event logs. The project’s objective is to develop a versatile and user-friendly library that incorporates multiple methods, including a Convolutional Neural Network (CNN) and a parameterizable heuristic approach, to accurately identify the Case ID column. By offering flexibility to users, they can choose individual methods or a combination of methods based on their specific requirements, along with adjusting heuristic-based formula coefficients and settings for fine-tuning the identification process. This report presents a comprehensive exploration of related work, methodology, data understanding, methods for Case ID column identification, software library development, and experimental results. The results demonstrate the effectiveness of the proposed methods and their implications for decision support systems.