Open Source AI-Enhanced 3D-Printed Insulin Pump

This research aims to develop functional 3D printed low-cost wearable insulin pumps for precise and personalized dosing for type 1 diabetes mellitus (T1DM) patients. The system utilizes artificial intelligence (AI) insulin to predict personalized bolus insulin dosages and send the dosing configuration wirelessly to an in-house developed IoT-reconfigurable pumping system to deliver the specified dosage to the patient. Historical patients’ data are fed to a cloud-based machine learning model to predict the personalized amount of bolus insulin based on the patient’s blood glucose levels, the amount of carbohydrates consumed, and physical activity. The system delivers U100 insulin bolus dosage at a flow rate of $750 \mu \mathrm{l} /\min$ against an estimated 0.764 kPa backpressure produced by the patient’s subcutaneous tissue, which is 6 to 8 times faster than current insulin pumping systems. In addition to the system’s AI-enhancement, IoT connectivity, small footprint, and low power consumption, the system’s cost is predicted at under ${\$}$35, making it an excellent candidate to fill a market gap and act as a fully autonomous artificial pancreas that improves the quality of life of T1DM patients.