mHealth-benaderingen in Sub-Sahara Afrika en gebruiksgericht ontwerp van een mHealth-app voor zelfmanagement van diabetes

Diabetes mellitus has become one of the most pressing global health challenges of the twenty-first century, with prevalence rising most steeply in low and middle-income countries. Sub-Sahara Africa (SSA) is experiencing rapid demographic and lifestyle changes that contribute to the growing burden of both type 1 and type 2 diabetes, yet healthcare systems in the region remain severely under-resourced. Limited access to insulin, diagnostic tools and trained professionals exacerbates mortality and complications, while systemic weaknesses such as fragmented care delivery and poor patient education further undermine effective management. Against this backdrop, mobile health technologies offer a promising opportunity to expand access to care, empower patients in self-management and support overstretched health systems. This dissertation investigates the landscape of mobile health in SSA, develops an AI based blood glucose prediction model that performs well under sparse data availability, as well as a user-centered smartphone app for diabetes self-management. The first stage of the research consisted of a comprehensive mapping of the evidence on existing mobile health approaches in SSA. A scoping review revealed that, while numerous interventions have been introduced across the region, most are confined to pilot or proof-of-concept stages. SMS text-based systems remain the most widely used, with applications in HIV, maternal and child health and cancer, whereas more advanced app-based and wearable sensor solutions are far less common. Telemedicine solutions are emerging, often with larger-scale implementation, but diabetes receives far less attention than other conditions. A systematic comparison between SSA and Europe highlighted stark contrasts: while European systems rely heavily on apps, wearables and sensors to support monitoring in ageing populations and chronic diseases such as cardiovascular conditions and diabetes, SSA depends on basic SMS and call-based tools focused primarily on young mothers and infectious diseases. These differences reflect disparities in demographics, infrastructure and health system capacity, underlining the need for contextualized digital health strategies in Africa. The dissertation also examined the broader health and demographic context of both regions. SSA is at the intersection of stage two and three of the demographic transition model, with a young population, high fertility rates and significant mortality driven by non-communicable, communicable, maternal and neonatal conditions. Europe by contrast is situated between stages four and five, with low fertility, ageing populations and a disease burden dominated by non-communicable diseases. Importantly, this research underscores that the dual burden of communicable and non-communicable diseases in SSA requires a more balanced approach, with far greater investment in digital strategies for diabetes care. A key component of the dissertation was a qualitative study conducted in Ethiopia to assess barriers and enablers of mobile health adoption from the perspectives of patients, physicians and healthcare executives. A total of 137 determinants were identified, comprising 68 barriers and 69 enablers. Patients reported low awareness of mobile health solutions, affordability challenges and concerns about trustworthiness as major obstacles, while also expressing strong interest in user-friendly tools that could automate monitoring and provide guidance. Physicians highlighted the lack of established digital systems, limited training, concerns about patients' digital literacy and infrastructural weaknesses such as unreliable internet and electricity. At the same time, they recognized the potential of mobile health to reduce workload, improve information management and enhance quality of care. Healthcare executives acknowledged competing priorities in under-funded health systems, high start-up costs and the non-sustainability of donor-led projects as persistent barriers. Yet they also emphasized the growing government commitment to digitalization, the availability of educated technology professionals and the potential of mobile health to extend services into remote and underserved communities. These findings reinforce the importance of designing interventions that are contextually relevant, affordable and embedded within health system strategies rather than isolated pilot projects. The dissertation also explored the lived experiences of people with diabetes in Ethiopia to identify user needs for self-management. Patients described multiple challenges, ranging from insulin shortages, poor access to glucose monitoring devices and financial burdens, to psychosocial barriers such as stigma, emotional distress and lack of family or community support. They emphasized the need for reliable and affordable glucose meters, improved insulin delivery mechanisms and integrated guiding systems or apps that could provide education, reminders and monitoring functions. Beyond technological solutions, patients also called for better healthcare services, customized treatment plans, social awareness and emotional support networks. This user-centered understanding of diabetes self-management informed the design requirements for the subsequent technological development. To address these needs, a deep learning model was developed for blood glucose prediction using the Ohio T1DM public dataset. A Gated Recurrent Unit (GRU) model was trained with input features such as glucose measurement, insulin units, meal value and indicators and evaluated on down-sampled sparse dataset to reflect the realities of SSA populations where continuous monitoring devices are scarce. The model demonstrated strong predictive performance across different down-sampling intervals and prediction horizons, with potential clinical relevance for proactive management of blood glucose levels. Additionally, the dissertation employed a user-centered design methodology to co-develop and test a prototype mHealth application called DiabAid Nexus. Patients and healthcare providers were actively engaged throughout the design process to ensure cultural and contextual relevance. Usability testing conducted in Ethiopia confirmed the platform's acceptability and highlighted its potential to support diabetes self-management by providing dashboard for glucose and insulin logging, trend analysis, educational resources, reminders and monitoring tools in both Amharic and English. The findings of this PhD dissertation have several important implications. First, they show that digital health strategies in SSA must be grounded in user needs and local realities to ensure adoption and sustainability. Second, they demonstrate the potential of artificial intelligence to be adapted for data-scarce environments. Third, they highlight that mobile health interventions hold significant potential for improving diabetes care and that user-centered design ensures that the app is closely aligned with the needs and realities of the target population. In conclusion, this dissertation delivers both a conceptual and practical contribution through the development of an artificial intelligence model and mobile health prototype tailored for SSA. By situating technological development within the lived realities of patients and the systemic constraints of low-resource settings, the research demonstrates a pathway toward equitable, scalable and contextually grounded digital solutions for diabetes care. The proposed DiabAid Nexus app exemplifies how user-centered approaches can transform digital innovation into a meaningful tool for improving self-management and ultimately reducing the burden of diabetes complications in SSA.