Revolutionizing wound care with AI

As a surgeon and researcher at the University of Arizona College of Medicine – Tucson, Dr. Geoffrey Gurtner was surprised to find a significant gap in medical research: no one was continuously monitoring wounds to catch complications early.

The revelation led him to take a bold leap, developing artificial intelligence-enabled bandages to transform patient care. Gurtner, professor and chair of the Department of Surgery, envisions “smart bandages” that use AI to predict infections before they affect wound healing.

In 2023, his team published a paper in Nature Biotechnology that showed a flexible, sensor-laden bandage could track and treat wounds in mice. With Department of Defense funding support, Gurtner’s team is now advancing human clinical trials to predict infections early, reduce health care costs and improve outcomes for patients.

Gurtner sat down with the U of A Health Sciences Office of Communications to discuss the progress, surprises and future of this groundbreaking work.

Q: What inspired you to develop AI-enabled smart bandages for wound monitoring?

Dr. Gurtner: The project started at Stanford when my collaborator, Zhenan Bao, a flexible electronics chemical engineer who makes soft and flexible electronics, reached out to see if we could work together. We saw a gap – nobody was really monitoring surgical sites or wounds continuously, unlike consumer wearables such as Apple or Google watches that track heart rate from the wrist. I was surprised to find no existing data or publications on this for such a common issue like wounds. We thought, let’s start by exploring commercially available, handheld sensors approved by the Food and Drug Administration to see if we could predict complications in wounds. It was a bit of a flyer, but if this idea worked, we could create clinical wearables to decentralize health care and catch problems early. Once we saw it was doable, it became something we couldn’t stop pursuing.

Q: What is your current clinical trial investigating?

Dr. Gurtner: Right now, we’re monitoring wound patients at Banner – University Medical Center Tucson, where we see many patients with diabetic foot ulcers, a major cause of nontraumatic amputations. People with diabetes tend not to heal well, and foot ulcers can lead to a downward spiral with decreases mobility and even mortality. We’re also looking at venous leg ulcers, another major type of wound, and burn patients. These are common, especially in Tucson, where we have lots of patients with diabetes, including large Hispanic and Native American populations that are disproportionally affected.

We’re past the prototype stage and have enrolled 83 patients in a trial looking at handheld sensors and how they correlate to amputations and other clinical outcomes. Currently, we see patients once a week for sensor readings in the office. Even with this sparse data set – one moment every 168 hours – we’ve found trends that correlate with infections or deterioration before they’re clinically diagnosed. In the future, we’d like to monitor patients who’ve had invasive surgical procedures, such as knee replacements, to predict wound infections before they develop into full-blown infections.

Q: What is the next step for the smart bandage?

Dr. Gurtner: We are working on a wearable, continuous monitoring device. We’re optimistic that with 168 hours of continuous data, we’ll have an enormous data set that will increase our resolution and predictive ability for bad outcomes.

Q: How are you using AI and data to refine the smart bandage’s capabilities?

Dr. Gurtner: Our model is unbiased, starting with nine commercially available sensors: temperature, humidity, pH, hemoglobin oxygenation, impedance and others. We correlated this data with electronic medical records to see which patients needed amputations or had complications. This was a big correlation exercise with lots of data. Even measuring once a week, we found a 90% positive predictive value for complications. Surprisingly, we only needed two sensors, not all nine, which makes a wearable device feasible, since a nine-sensor device would be very complicated. Now, with continuous monitoring for 168 hours, we’ll generate terabytes or petabytes of data, which only AI can handle. We’re refining an algorithm to say, ‘This wound is infected’ or ‘This wound will become infected,’ and trigger treatments like electrical stimulation, which would happen autonomously and not require a visit to the doctor. The electronic medical record data helps us refine these predictions quickly to deliver actionable outcomes.

Q: What surprising insights have you gained about wound healing through this research?

Dr. Gurtner: One obvious finding is that as a wound heals, skin temperature goes down. But a big surprise was pH. There’s an enormous amount of literature about pH, lactate and the acid-base equilibrium of wounds being critical, but we found no relationship between pH changes and wound outcomes. That’s very surprising and challenges what a lot of people thought was important. These insights are shaping how we think about wound healing.

Q: What is the timeline for bringing these smart bandages to clinical use?

Dr. Gurtner: We’ll start applying the continuous bandages this fall, but there are challenges to troubleshoot, like data transmission. Continuous Wi-Fi or Bluetooth drains the battery in 10 hours, so we’re creating data packages downloaded every 6 to 8 hours to last a week, since patients come in weekly. Our Department of Defense grant includes funding for FDA approval. After that, a startup company might manufacture the device. Once we prove this simple device can predict outcomes as an early warning system, I think larger medical device companies will become interested.

Q: Do you have other research plans for the smart bandage?

We plan to extend this to post-surgical monitoring, where surgical infections are a big problem, to predict readmissions and complications. We’ve also applied for a grant to explore new sensing modalities and actuators to treat infections autonomously, so the bandage could treat an infection without a physician’s involvement. I think this is where medicine will be in the next 50 years.