Wearable AI Sensor Supports Personalized Health Data Processing, Analysis
Researchers at the University of Chicago have created a wearable computing chip, which can analyze a person’s health data in real time using artificial intelligence.
Source: Getty Images
By - New research from the University of Chicago’s Pritzker School of Molecular Engineering (PME) shows that a flexible, stretchable computing chip worn directly on the skin can collect and analyze health data in real time using artificial intelligence (AI).
The device, known as a neuromorphic computing chip, uses semiconductors and electrochemical transistors to collect data from biosensors touching the skin. Unlike other wearables, such as smartwatches, which leave a small gap between the user’s skin and the device, the chip is designed to be worn directly on the skin to improve sensor accuracy and data collection.
To achieve this, the researchers needed to accommodate the movement of the wearer’s skin, which provides a unique challenge for wearable device developers. For the chip to integrate seamlessly with the skin, it had to stretch and move similarly. To give the chip the necessary flexibility, the researchers utilized polymers, which can not only stretch and bend but can also be used to build transistors and semiconductors.
However, these improvements in the chip’s ability to conform to the skin and collect better data resulted in a need for additional processing power. The more data a wearable can collect, the more complex the data analysis becomes. Smartphones, which are often connected to wearable technology to send and process data, are not capable of doing so for a device like a neuromorphic computing chip.
To address this, the research team utilized AI algorithms within the chip to help efficiently analyze relevant patterns within the data.
After designing the device, the researchers tested its utility as a health data analytics tool. The team began by training the device’s machine-learning (ML) algorithm to analyze and classify electrocardiogram (ECG) data into five categories: one for healthy ECG signals and four for unhealthy or abnormal signals.
After the device was trained, they validated it using new ECG data it had never seen before. The device was able to accurately classify the ECGs. Further, the researchers found that it could do so even when stretched or bent, as it would be when worn by a user.
The researchers note in the press release that this is just the beginning of their research in this area. They hope that this is a step toward a future where diseases can be easily detected before symptoms appear using wearables and continuous health data tracking.
They also said that such a device has the potential to be used like many implantable medical devices.
“If you can get real-time information on blood pressure, for instance, this device could very intelligently make decisions about when to adjust the patient’s blood pressure medication levels,” said Sihong Wang, PhD, assistant professor of molecular engineering at PME, in the press release. Similar techniques are already being used in some devices, such as implantable insulin pumps.
“Integration of artificial intelligence with wearable electronics is becoming a very active landscape,” Wang continued.
The research team is currently planning new iterations of the chip, which will expand the types of devices it can integrate with and the ML algorithms it uses.
