Artificial intelligence and algorithmic clinical decision support at scale in African health systems: a living scoping review and evidence and gap map

Objective: to map AI and algorithmic clinical decision support deployments that have reached material scale in routine care in African health systems, and to characterise the record by technology class, failure mode, evidence characteristics, matched process-patient outcome profiles, value evidence and sustainment. Such systems are in routine use at scale in a number of African countries. The evidence about what has been measured at scale, with what results and at what cost, has not been assembled on a common deployment unit. This review builds one. Inclusion criteria: systems that take patient-specific inputs and produce or prioritise a clinical judgement (diagnosis, classification, risk, triage category, management or treatment recommendation), deployed in routine care in any African health system, meeting at least one of three scale routes: ten or more health facilities in routine use; documented national or sub-national government adoption with dated evidence of routine use; or 10,000 or more documented routine consultations, encounters or images. Deployments from 1 January 2015 to the search date. Any evidence type and any language. Methods: reported per PRISMA-ScR and JBI scoping review methodology, with the consultation stage from Levac, Colquhoun and O'Brien. Searches run across eight bibliographic databases, preprint servers, normative and ministry sources in English, French, Portuguese and Arabic, conference proceedings and citation chasing. Dual independent screening with an evaluated AI third arm. Charting on linked deployment and report forms against a deposited codebook. Appraisal routed by estimand. Descriptive synthesis, including a matched process-patient outcome analysis and a value-evidence distribution, presented as a Campbell-standard living evidence and gap map, updated annually for at least three years. Expected outcomes: a failure-mode taxonomy for AI and algorithmic clinical decision support deployment in African health systems; an evidence and gap map; and a value-evidence distribution identifying where cost, comparative economic evaluation and budget-impact evidence is present or absent for deployments that have reached scale.