2024 Proceedings
Permanent URI for this collection
Browse
Recent Submissions
Now showing 1 - 5 of 24
- ItemOpen AccessImproving the automated estimation of malaria parasite density detection(MIRG, 2024-11-25) Ibrahim, L.M.C; Ismail, L.; Theopila, D.; Jules, D; Habiboulaye, A.B.Malaria is an extremely deadly disease caused by the Plasmodium parasite, which is transmitted to humans through the bites of infected mosquitoes. The fight against this disease relies on early and accurate diagnosis. However, existing diagnostic tools have significant limitations, which are aggravated by a growing lack of qualified medical professionals globally, making the management of malaria more complicated than ever. Computer vision techniques have been largely used in this domain to propose approaches for combating malaria. While these approaches offer various advantages, they also present certain limitations. In this article, we have collaborated with malaria diagnostic experts to develop an innovative malaria diagnostic system. This system is based on the Yolov5 object detection model, that we modified to improve the precision of detection of small objects, such as malaria parasites, a particularly complex challenge in the medical field. This approach enabled the simultaneous identification of parasites and white blood cells (WBC), offering a more complete analysis than most previous research. After training, the model achieved 93.2% precision, 96.6% recall, 98.1% mAP50 and 94.86% F1 score for malaria parasite detection. For the WBC detection, performance was even higher, with 99.7% precision, 99.6% recall, 99.5% mAP50 and 99.64% F1 score. We have consequently used this model to develop our precise diagnostic system, which is also accessible to non-experts, enabling us to respond effectively to the growing shortage of qualified medical practitioners. The system has been tested on a set of 19 blood smear slides, and the results confirm that our approach offers a malaria diagnostic precision comparable to that of an expert.
- ItemOpen AccessEthical Considerations in Using AI for Mental Health Diagnosis and Treatment Planning: A Scoping Review(MIRG, 2024-11-25) Ojo, YIntegrating Artificial Intelligence (AI) with mental healthcare presents a paradigm shift in diagnosis and treatment planning, offering potential efficiency, accuracy, and personalisation improvements. However, this technological advancement allows for the exploration of a complex array of ethical challenges that demand careful consideration. This research explores the vital ethical dimensions surrounding the adoption of AI in mental health contexts, emphasising the reason for a balanced approach that maximises benefits while mitigating risks. Central to these considerations is the imperative of privacy and data protection. This type of mental health information requires comprehensive robust safeguards to prevent unauthorised access or misuse while allowing for responsible data utilisation to drive AI-powered advancements. The assurance of fairness and non-discrimination in AI systems is critical, as racial bias could exacerbate disparities in mental healthcare access and outcomes. Transparency and explainability emerge as crucial factors in fostering trust and accountability. AI systems must be capable of providing clear rationales for their diagnostic and proposed treatment planning, which aids clinicians and patients to make informed decisions. This transparency is intimately linked to the principles of autonomy and informed onsent, requiring that individuals fully understand the role of AI in their treatment and have the agency to accept or decline its use. The integration of AI also necessitates a reevaluation of professional ethics and responsibilities for mental health practitioners. As AI systems assume more significant roles in diagnosis and treatment planning, the boundaries of professional judgment and accountability must be delineated. Moreover, the broader societal implications, including potential changes in public perception of mental healthcare and shifts in the healthcare workforce, warrant careful consideration. Regulatory and governance frameworks play a pivotal role in addressing these ethical challenges. Policymakers face the complex task of developing adaptive regulations that foster innovation while ensuring robust ethical safeguards. This requires a collaborative approach involving clinicians, researchers, ethicists, patients, and technology developers.
- ItemOpen AccessMachine Learning for Personalized Medicine in Sub-Saharan Africa: A review(MIRG, 2024-11-25) Majesty Akpara; Chika Yinka-BanjoPersonalized medicine (PM) has transformed healthcare globally by customizing treatments based on individual characteristics. Sub-Saharan Africa (SSA) faces pressing healthcare challenges, but machine learning can enhance personalized medicine by analyzing complex data efficiently. This technology improves treatment outcomes, diagnostics, and drug discovery. PM can significantly improve treatment outcomes, reduce adverse effects, and enhance diagnostics in SSA, where genetic diversity necessitates such tailored approaches. However, implementing ML in PM faces challenges, including data quality issues, infrastructure deficits, workforce shortages, and limited access to health information technology. Future directions for adopting PM in SSA involve strengthening health data systems, building capacity in data science and AI, and fostering public-private partnerships. Addressing these challenges and leveraging opportunities can improve healthcare outcomes, reduce costs, and advance pharmaceutical research in the region. Machine learning holds immense potential to enhance personalized medicine in SSA, promising a transformative impact on healthcare delivery and patient outcomes.
- ItemOpen AccessThe Future of Pharmaceutical Information: A Case for Generative AI Over Pharmacopoeias and Medical Apps(MIRG, 2024-11-25) Bashirudeen Opeyemi Ibrahim; Olubayo Adekanmbi; Anthony SoronnadiTraditional sources of pharmaceutical information, such as pharmacopoeias and medical apps, need more scope, updated frequency, and depth of information as Artificial intelligence (AI) evolves, particularly through generative AI and large language models (LLMs) and a unique opportunity exists to remodel how healthcare professionals access pharmaceutical data. This paper explores how generative AI can surpass traditional methods by providing up-to-date, personalised, comprehensive drug information. When trained on extensive pharmaceutical datasets from codices, pharmacopoeias, and regulatory bodies like the FDA and WHO, generative AI can generate novel insights and streamline access to current pharmaceutical knowledge. By incorporating real-time updates and query-based systems such as AI-powered chatbots, generative AI ensures healthcare professionals can retrieve more accurate, relevant, and personalised drug interactions, dosage forms, and side-effect profiles. Findings suggest that generative AI offers grreat advantages over traditional drug information, enhancing decision-making and patient care outcomes even though its adoption raises concerns about data privacy, bias, and the reliability of AI-generated content. Rigorous validation processes and continuous updates are essential to maintaining trust in the system as generative AI is a powerful supplement to traditional pharmaceutical information sources, facilitating better-informed decisions by healthcare professionals while addressing many of the inherent challenges of pharmacopoeias and apps. Generative AI can help shape the future of pharmaceutical care which eventually improves patient outcomes by upholding ethical standards, ensuring accuracy, and integrating feedback from medical experts.
- ItemOpen AccessA Recipe Recommender System Using Natural Lanuguage Processing and Similarity Measures(MIRG, 2024-11-25) Bukola Badeji-Ajisafe; Betsy G. Otoyo; Temiloluwa A. Adebola; Abimbola O. Ajibade; Olanike C. Akinduyite; Stephen. E ObamiyiPersonalized recipe recommendation has become a compelling need, driven by the motivation to develop intelligent systems that can match user preferences with the most related suggestions. This paper presents a recipe recommender system in which several word-embedding models—namely, Word2Vec, FastText, and Sentence Encoder (Mini LM)—are evaluated for obtaining high semantic similarity between user input and recipes in the dataset. Cosine similarity is used as the primary metric for measuring how close user inputs are to the existing recipes. The results show that the Word2Vec model performs best and has good understanding of language patterns related to the recommendation of recipes in a more precise manner. The study hence justifies using Word2Vec as an asset for future personalized recommendation systems.