Personalized Algorithms and Techniques for Development of a Smart Infusion Pump for ICU

In hospitals, intravenous infusion pumps are used for the delivery of medicines such as antibiotics, insulin, inotropes, vasopressors, and certain nutrients. Such smart infusion pumps can deliver the fluids in controlled amounts without any external aid. Currently, the type of medicine, number of medicines, infusion rate, and its dosage are administered in a clinically guided, yet generalized manner. For every person, the infusion pump delivers a preset amount of medicine without instantaneously monitoring and considering it’s effects on patients. The patients can be children, adults and geriatric with different diseases. For them the smart infusion devices are designed in such a way that a constant rate of medicine is delivered every hour, and the rate is manually changed in emergencies. The manual adjustment within an hour itself can cause an overdose or underdose of medicine even for a very short duration. This uncontrolled drug delivery may not be tolerated by ICU patients. In the case of post-operative patients, the vital parameters such as temperature, blood pressure, heart rate, oxygen saturation level etc. vary instantaneously. Most of them are supported with various drugs such as adrenaline, nor-adrenaline, vasopressin etc. The preset amount of infusion will not be suitable for them. Hence, we propose personalized algorithms and techniques for developing a smart infusion pump, which monitors the vital parameters of post-operative patients and uses them to control the delivery of medicine. Machine learning algorithms can be integrated into our architecture for personalize infusion of drugs with precise control of the stepper motor.