As a wound heals, it goes through several phases: Clotting to stop bleeding, immune system response, scab formation and scarring. A wearable device called “a-Heal”, developed by engineers at the University of California at Santa Cruz, aims to optimize each stage of the process. The system uses a tiny camera and AI to detect the stage of healing and deliver treatment in the form of medication or an electric field. The system responds to the patient’s individual healing process and provides personalized treatment. The wearable, wireless device could make wound therapy more accessible for patients in remote areas or with limited mobility. Initial preclinical results published in the journal “npj Biomedical Innovations” show that the device successfully accelerates the healing process.
Development of a-Heal
A team of researchers from UC Santa Cruz and UC Davis, funded by the DARPA-BETR program and led by Marco Rolandi, Baskin Engineering Endowed Chair and Professor of Electrical and Computer Engineering (ECE) at UC Santa Cruz, has developed a device that combines a camera, bioelectronics and AI to accelerate wound healing. The integration into one device makes it a “closed system” – one of the first of its kind for wound healing, to the researchers’ knowledge. “Our system picks up all signals from the body and optimizes the healing process with external interventions,” said Rolandi.
The device uses an integrated camera, developed by Mircea Teodorescu, also an associate professor of ECE, and described in a study in Communications Biology, to take photos of the wound every two hours. The photos are fed into a machine learning (ML) model developed by Marcella Gomez, Associate Professor of Applied Mathematics, which the researchers describe as an “AI doctor” running on a nearby computer. “Basically, it’s a microscope in a bandage,” says Teodorescu. “Individual images tell us little, but over time, through continuous imaging, the AI can identify trends, determine wound healing stages, report problems and suggest treatments.” The AI doctor uses the image to diagnose the wound stage and compares it to the stage the wound should be at according to a timeline for optimal wound healing. If the image shows a delay, the ML model applies a treatment: either a drug administered via bioelectronics or an electric field that can promote cell migration towards wound closure.
Accelerating Wound Healing
The treatment delivered topically via the device is fluoxetine, a selective serotonin reuptake inhibitor that regulates serotonin levels in the wound and improves healing by reducing inflammation and promoting wound closure. The dose, determined through preclinical studies by the Isseroff group at UC Davis to optimize healing, is delivered through bioelectronic actuators on the device developed by Rolandi. An electric field, optimized to enhance healing and developed through earlier work by Min Zhao and Roslyn Rivkah Isseroff at UC Davis, is also delivered through the device.
The AI physician determines the optimal dosage of medication to be administered and the strength of the electric field applied. After the therapy has been applied for a certain period of time, the camera takes another image and the process starts all over again. During use, the device transmits images and data such as the healing rate to a secure web interface so that a human doctor can manually intervene and fine-tune the treatment if necessary. The device attaches directly to a standard bandage for convenient and safe use. To evaluate the potential for clinical use, the UC Davis team tested the device in preclinical wound models. In these studies, wounds treated with a-Heal healed approximately 25% faster than with standard treatment. These results underscore the technology’s potential not only for accelerating the healing of acute wounds, but also for revitalizing the healing of chronic wounds.
Model of Reinforcement Learning
The AI model used for this system, developed under the direction of Marcella Gomez, Assistant Professor of Applied Mathematics, uses a reinforcement learning approach, described in a study in the journal Bioengineering, to mimic the diagnostic approach of physicians. Reinforcement learning is a technique in which a model is designed to fulfill a specific end goal and learns through trial and error how best to achieve that goal. In this context, the model is set the goal of minimizing the time to wound closure and is rewarded for progress in achieving this goal. It continuously learns from the patient and adapts its treatment approach.
The reinforcement learning model is driven by an algorithm developed by Gomez and her students called Deep Mapper. This algorithm, which is described in a preliminary study, processes wound images to quantify healing progress compared to normal and map it along the healing trajectory. Over the time that the device is applied to a wound, it learns a linear dynamic model of healing to date and uses this to predict further healing progression.
It is not enough just to have the image, you also have to process it and put it into context. Then you can apply feedback control,” said Gomez. This technique allows the algorithm to learn in real time the effects of the drug or electric field on healing, and guides the reinforcement learning model’s iterative decision making on how to adjust the drug concentration or electric field strength. The research team is currently investigating the potential of this device to improve the healing of chronic and infected wounds.