Determinants of collaboration dynamics and challenges in North-South data-driven health research in the African context: an epistemic injustice perspective.

INTRODUCTION: Collaborations between Global North and South institutions are increasingly important for advancing artificial intelligence (AI) and data science in healthcare, offering opportunities to address context-specific challenges and reduce inequities. However, these partnerships often face persistent issues related to sustainability, equity, and long-term impact. Drawing on concepts of epistemic injustice and structural imbalances between partners, this study examines the factors shaping the effectiveness of North-South data-driven health collaborations, with particular attention to which factors exert the strongest statistically significant influence on collaboration effectiveness. METHODS: The study conducted a cross-sectional survey of participants from the Data Science Initiative for Africa (DS-I Africa) network, covering the entire continent of Africa, and their partners from the Global North, and applies structural equation modeling to assess key determinants of collaboration outcomes. RESULTS: The analysis focuses on governance and leadership, human capacity, and project and institutional practices. Human capacity is the strongest positive predictor of collaboration effectiveness (β = 0.467, p = 0.008). Project and institutional practices show a significant negative association with collaboration effectiveness (β = -0.464, p = 0.016). Governance and leadership also demonstrate a significant effect on collaboration outcomes. DISCUSSION: The findings demonstrate that sustainable and equitable North-South collaborations depend on the alignment of governance, human capacity, and institutional practices. Moreover, addressing epistemic injustice requires system-level changes that promote shared leadership, ethical data governance, and strengthening human capacity as a reciprocal and inclusive process central to achieving balanced knowledge production and long-term collaboration impact.