Authors: Abdel Nacer Amoukou Issaka, Amadou Bouba Traore Hassane, Ibrahim Alkassoum, Zeinabou Tondi Moussa, Abdou Atté Bachir, Djibrilla Baoua Bonkano, Djongolé Hassane Moussa, Roukaya Karimou Issaka
Abstract
Introduction: The digital transformation of healthcare systems has been accompanied, since the early 2020s, by the rapid adoption of artificial intelligence (AI), further accelerated by the emergence of generative AI and large language models (LLMs). Beyond broad discussions about the potential of AI, healthcare professionals, researchers, and educators need practical guidance regarding available tools, their applications, performance, limitations, and the conditions required for their responsible use.
Objective: To provide an international, critical, and educational synthesis of the major applications of AI in clinical practice, biomedical research, and health sciences education, while examining its potential benefits, limitations, associated risks, and the conditions required for safe and responsible use, particularly in resource limited settings.
Methods: A structured narrative review of the literature was conducted using publications indexed in PubMed/MEDLINE, Scopus, Web of Science, and the Cochrane Library. The literature search was supplemented by institutional documents from the World Health Organization (WHO), UNESCO, and the European Union, as well as selected systematic reviews, evidence syntheses, and regional reports addressing the use of AI in healthcare in sub Saharan Africa. Publications primarily published between 2020 and 2026 were prioritized.
Results: Available evidence demonstrates a rapid diversification of AI applications in healthcare. Generative AI systems and LLMs can support medical information synthesis and interpretation, clinical decision support, literature searching, scientific writing and communication, development of educational materials, clinical simulation, and selected epidemiological surveillance activities. These technologies may improve access to information and increase the efficiency of selected healthcare and research tasks. However, their use is associated with important risks, including hallucinations and factual errors, algorithmic bias, potential breaches of data privacy, limited traceability and accountability, and the risk of cognitive dependence among users. These challenges are particularly relevant in healthcare systems with limited human, digital, and regulatory resources.
Perspectives: Between 2026 and 2030, advances in multimodal AI, agentic systems, and personalized medicine may substantially expand the scope of AI applications in healthcare. These developments will require robust governance frameworks, scientific validation, human oversight, data protection, and appropriate professional training. Particular attention should be given to the linguistic, cultural, technological, and socioeconomic adaptation of AI solutions to resource limited settings to prevent the widening of existing digital and health inequities.
Conclusion: The key challenge is no longer whether AI will be used in healthcare, but how it can be implemented safely, ethically, scientifically, and in alignment with the needs of healthcare professionals and populations. In low and middle income countries, particularly in the Sahel, AI integration should be approached as a gradual process based on rigorous validation, human oversight, capacity building, and context specific governance.
Keywords: Artificial intelligence; machine learning; generative artificial intelligence; large language models; clinical decision support; ethics; data privacy; medical education; biomedical research; sub Saharan Africa.