Planning a Clinical trial for a Stroke Diagnostic Device in a Resource-Constrained Health System: A Case Study of the StrokeFinder MD100 in Rwanda

Abstract

Stroke is the leading cause of death in Rwanda, and its incidence has grown sharply over recent decades. Yet confirmatory brain imaging is only available at four tertiary hospitals in the country, and most patients arrive too late to receive the only clot-dissolving treatment available. The Medfield StrokeFinder MD100 is a portable device that uses microwave technology to detect stroke at the bedside, without requiring a CT scanner. It has shown promising results in European and Australian studies, but has never been tested in sub-Saharan Africa, where patients, disease patterns, and care pathways differ considerably. This thesis develops a planning framework for a GCP-compliant clinical validation study of the MD100 in Rwanda. The framework is based on sixteen interviews conducted in Kigali between March and April 2026 with clinicians, hospital researchers, national regulatory bodies, and a trial consultancy. Analysis of the roughly 80,000-word interview corpus identified three key findings: the locations where the device would be most useful are also the hardest places to run a validation study; clinical trial terminology is not consistently understood across the healthcare workforce; and knowledge of AI-specific regulation is concentrated at a single institution. These findings are translated into a four-phase trial-planning framework. The core of the framework is a two-stage trial: A first stage at tertiary hospitals establishes how accurately the device classifies stroke against CT imaging under controlled conditions. A second stage at district hospitals then measures whether deploying the device in routine care actually speeds up the time from first clinical contact to diagnosis.