Abstract
Infections in wound care are extremely difficult to detect early, leading to increased risk of complications, reinfection, and hospital readmissions. To address this gap caused by the lack of continuous monitoring, Phong Le, my technical project collaborator, and I developed a wound monitoring system that analyzes physiological changes at the wound site and identifies early signs of infection risk. Rather than relying on a trained machine learning model, the system uses a rule-based temperature heuristic grounded in clinical research on wound temperature and infection patterns. However, this technical solution raises a broader question: how can trust be established in AI-driven healthcare systems as they continue to expand opportunities to improve patient care? Engineers designing these systems must consider human and ethical factors rather than focusing solely on technical performance.
My STS research was focused on trust in AI-driven healthcare systems. I found that patients and healthcare workers are often hesitant to trust AI in healthcare in high-stakes medical situations, particularly when they lack understanding of how these systems operate. Trust increases when systems are transparent and designed to support clinicians rather than replace them. In contrast, trust decreases when patients are unaware that AI is being used or when the system’s decision-making process is not clearly communicated. For healthcare to continue evolving, several key ethical issues must be addressed, such as ensuring informed consent, reducing algorithmic bias, and preventing overreliance on AI at the expense of human judgment. While AI has the potential to significantly transform healthcare, its effectiveness depends on patient acceptance and transparency.
The technical portion of my research produced a wound monitoring prototype that continuously monitors wound conditions using high-precision temperature sensors integrated into a reusable capsule that can be attached to bandages. The sensor transmits data over Wi-Fi to a backend server, where a dashboard allows healthcare workers to observe real-time trends and receive early infection alerts. A rule-based heuristic analyzes temperature readings over time, identifying patterns such as drops below the baseline or rapid rebounds above it to predict infection risk. This approach allows users to clearly understand why an alert was triggered, prioritizing transparency. While the system does not rely on machine learning models, it serves as a foundation for future systems that could incorporate machine learning as reliable datasets become available.
The next major advancements in healthcare will depend on artificial intelligence. However, artificial intelligence should not be viewed as a replacement for healthcare professionals, but rather as a tool to support and enhance their skills. My STS research emphasized the importance of transparency in building trust, which directly influenced the design of the wound monitoring system by prioritizing interpretable, rule-based alerts that clearly communicate infection risk. Smart bandages, for example, have the potential to significantly reduce complications from surgical wounds, with even greater impact as more advanced methods, such as machine learning and artificial intelligence, are integrated into its design. For these technologies to be trusted by both clinicians and patients, they must remain transparent and follow ethical standards. While widespread adoption may not occur immediately or perform perfectly in early stages, society must begin to adapt to and integrate these technologies if we aim to continue making significant progress in healthcare.
