Low-Cost Insulin Pump with Predictive-Based Mitigation of Hyperglycemia and Hypoglycemia

Abstract

 Equitable access to insulin pumps, the gold standard for diabetes management, remains elusive in low- and middle-income countries (LMICs). For example, in countries like Ethiopia, fewer than 1% of insulin-dependent patients access this technology, hindered by prohibitive pricing and regulatory complexities. The current study proposes a low-cost insulin pump integrating custom-made hardware design with predictive software algorithms. The proposed design features a micro control unit (MCU) within a threaded spindle-based mechanical framework, housed in a 3D-printed casing. A complete prototype using an Arduino microcontroller has been developed, and testing was conducted using a vision-based syringe displacement method. Novel contributions of this design include a Kalman filter algorithm for hypoglycemia prediction, Bluetooth-enabled Android app control, and automation software delivering personalized insulin doses. This design significantly lowers the total cost of production, with material costs alone reduced to USD $95, while achieving accuracy approaching that of commercial pumps (84.7% of basal doses within ±5%) and a 3.49% average infusion error. By leveraging custom PCB and mechanical designs, it bridges frugal engineering with clinical functionality, offering scalability potential through local manufacturing. Future steps involve conducting IEC-aligned safety testing and benchmarking against commercial pumps (e.g., Medtronic 640G) to validate real-world performance. This innovation demonstrates a pathway to democratize advanced diabetes care in LMICs, combining cost-effectiveness (projected retail price of USD $285– $475, assuming 3 to 5 times the bill of materials cost - BoM) with regulatory-compliant safety. It has the potential to reduce complications and improve outcomes in resource-constrained settings.