ISEP - Dissertações de Mestrado
Permanent URI for this community
Browse
Browsing ISEP - Dissertações de Mestrado by Sustainable Development Goals (SDG) "03:Saúde de Qualidade"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- Immune profiling of a peptide/hyaluronan injectable matrix for intervertebral disc therapyPublication . MAGALHÃES, INÊS RIBEIRO; Ribeiro, Maria Cristina CastroLumbar disc herniations (LDH) are a leading cause of chronic low back pain, resulting from intervertebral disc degeneration. Conventional treatments focus on symptom relief and are often associated with high recurrence rates and long-term complications. In this context, regenerative medicine offers alternative strategies that aim to restore tissue function and promote healing. This study aimed to evaluate the immune response of human monocyte-derived macrophages in contact with hyaluronic acid (HA)-based injectable matrices. Human monocytes were isolated from healthy donors, differentiated into M0 macrophages, and polarized into M1 (proinflammatory) and M2 (anti-inflammatory) phenotypes. These cells were put in culture with various biomaterials, including HA of different molecular weights, peptide-functionalized HA, and peptides synthetised. Biocompatibility, viability, proliferation, and macrophage phenotype were assessed through fluorescence imaging, cytometry, and ELISA. Results showed that the tested biomaterials were non-cytotoxic and influenced macrophage behaviour by promoting regenerative profiles. This work supports the potential of peptide/HA-based biomaterials as injectable immunomodulatory systems for intervertebral disc repair, offering promising prospects for minimally invasive therapies targeting LDH.
- Systematic review of exploration strategies and evaluation metrics in reinforcement learning using an automated software testing solutionPublication . BERTÃO, AFONSO MEIRELES; Pereira, Isabel Cecília Correia da Silva Praça GomesIn an era defined by rapid technological evolution, software systems underpin essential domains such as healthcare, finance and digital communication. Central to these systems are REST APIs, which enable seamless interaction across distributed services. However, the increasing complexity and ubiquity of these APIs have elevated concerns surrounding their security and robustness. Vulnerabilities in REST APIs can result in severe consequences, including data breaches, service outages and compromised user trust highlighting the need for intelligent and adaptive testing solutions. This thesis explores the integration of reinforcement learning (RL) into automated software testing, with a focus on algorithmic exploration strategies and evaluation metrics applied within FuzzTheRest, a state-of-the-art REST API fuzzing tool. While FuzzTheRest leverages the Epsilon- Greedy exploration strategy for action selection, its simplistic balance between exploration and exploitation limits its effectiveness in navigating complex API environments and uncovering subtle security flaws. To address these limitations, this review investigates the potential of Boltzmann Exploration, a probabilistic approach that adjusts action selection based on a softmax distribution over Q-values, enabling more nuanced and informed exploration of the test space. The study conducts a comparative analysis of multiple input generation strategies including random, evolutionary and RL-driven techniques evaluating their performance in adaptation to APIs and endpoints with complex parameterization terms, successful total number of HTTP responses, exploration fuzzing metrics and computational efficiency. In particular, the review highlights the impact of exploration strategies like Epsilon-Greedy and Boltzmann on the quality and depth of fuzzing outcomes. By synthesizing current literature and experimental insights, this review lays the groundwork for advancing RL-guided fuzzing methodologies. The findings aim to empower developers, testers and security practitioners with a deeper understanding of how exploration strategies choices influence the effectiveness of automated testing tools, contributing to more secure and resilient software systems.