Digital communication for maternal and newborn health in sub-Saharan Africa: evidence, implementation and AI implications

Abstract

Digital health interventions are increasingly used to communicate antenatal, intrapartum, postnatal and newborn-care information in sub-Saharan Africa, but their effects depend on more than message delivery. This narrative review examines what recent evidence shows about the mechanisms, implementation conditions and outcomes of patient-facing and provider-facing digital interventions, with particular attention to rural and semi-rural settings. Peer-reviewed studies and authoritative guidance published from 2000 to 2026 were considered. Evidence was organised along a pathway from access and information delivery to comprehension, engagement, behavioural intention, care-seeking, service use, quality of care and clinical outcomes. Digital interventions can improve selected intermediate outcomes, including knowledge, birth preparedness, antenatal attendance, postnatal contact and newborn-care practices. Effects are heterogeneous, and positive findings are concentrated in specific contexts and outcome domains. Stronger signals are reported when digital communication is interactive, linked to a clinical helpdesk, extended to social supporters, or integrated into provider workflows. However, direct comparisons do not establish that human-mediated delivery is inherently superior to self-directed delivery. Device access, network and power reliability, cost, literacy, language, gendered phone control, privacy, trust, staff workload and health-system readiness repeatedly shape reach and implementation. Evidence for maternal or neonatal morbidity and mortality remains limited. No reviewed study directly evaluated a large language model as a maternal-health intervention. Existing evidence supports design requirements for future artificial-intelligence research, including local-language evaluation, clinical factuality testing, human escalation, privacy protection and implementation assessment, but does not support claims of effectiveness or readiness for deployment. Digital exposure should therefore be treated as an input to a care pathway, not as a clinical outcome.