Computer Control of Drug Delivery

Overview

This study was the first to demonstrate the potential of computer-controlled drug delivery for highly potent critical care medications, specifically targeting applications where avoiding fluid overload is crucial (e.g., pediatrics). Very young, sick children can be so small that even a tiny difference in fluid delivered in clinical settings can stress their organs and circulatory systems. As another example, the circulatory systems of those with heart failure often can’t accommodate any extra load.

Despite the critical need for precise drug delivery in these scenarios, intravenous infusions via syringe pumps often result in substantial delays in achieving target drug levels, especially at low flow rates. To address this issue, we developed a mathematical model of drug flow, creating visualizations to illustrate the unseen dynamics of drug delivery to clinicians. Our mathematical models used numerical approximations of fluid diffusion to model how drug mixes as it moves from the syringe pump through the tubing to the patient.

These models were then translated into control algorithms that used predicted drug delivery to reduce delivery delays by up to 78%. We also validated the model’s effectiveness through in vivo experiments. This work led to patents, and subsequent research is incorporating these algorithms into commercial infusion pumps. This research has the potential for high impact on patient safety and for efficacy of intravenous infusions based on the ability to deliver crucial medications in a timely fashion in critical care scenarios.

My Reflections on This Work

This project allowed me to draw on different aspects of my background in education, research, and applied problem-solving. I designed the drug flow model, based on the physics of diffusion, and developed visualizations that provided clinicians with a clearer mental model of drug delivery, making discrepancies between intended and actual delivery more apparent. It was an enjoyable challenge making the work more understandable and higher impact by also creating companion visualizations that made the unseen seen for a less technical, more clinical audience. Turning the mathematical models into control algorithms was a key step in the project and contributed to our patents. I also wrote software that translated the algorithms’ output into instructions to control the pumps and worked on laboratory and in vivo studies to validate the system’s effectiveness. This work gave me a chance to integrate technical modeling with practical application, and it reinforced the value of combining analytical and hands-on approaches to address real-world challenges in patient safety and treatment outcomes.