According to the developers, the device uses significantly less energy than current technologies, as the nanoelectronic technology can crunch large amounts of data and perform AI tasks in real time without sending data to the cloud for analysis.
With its footprint, low power consumption and a proven lack of lag time to receive analyses, the device can be incorporated into wearable electronics, like smart watches and fitness trackers, for real-time data processing and near-instant diagnostics.
To test the technology, researchers trained the device to interpret data from ECGs, a task that typically requires significant time from health care workers. The AI was then asked to classify six types of heart beats, such as normal to atrial premature beat and premature ventricular contraction.
It was found that the device could correctly identify an irregular heartbeat and determine the arrhythmia subtype from among six different categories with near 95 per cent accuracy.
The study’s senior author, Mark C. Hersam, said: “Today, most sensors collect data and then send it to the cloud, where the analysis occurs on energy-hungry servers before the results are finally sent back to the user. This approach is incredibly expensive, consumes significant energy and adds a time delay.
“Our device is so energy efficient that it can be deployed directly in wearable electronics for real-time detection and data processing, enabling more rapid intervention for health emergencies.”
According to the engineers, whilst current silicon-based technologies need over 100 transistors to categorise large datasets, Northwestern’s nanoelectronic device can perform the same machine-learning classification with two nano-transistors that are dynamic enough to switch among various steps of data processing.
Hersam stated that the aim is to incorporate these nanoelectronic devices into everyday wearables, personalised to each user’s health profile for real-time applications. This would enable people to make the most of the data they already collect without sapping power, as well as localising their information to ensure protection and privacy.
“Artificial intelligence tools are consuming an increasing fraction of the power grid,” Hersam said in a statement. “It is an unsustainable path if we continue relying on conventional computer hardware.”
The study, “Reconfigurable mixed-kernel heterojunction transistors for personalized support vector machine classification,” was supported by the US Department of Energy, National Science Foundation and Army Research Office, and can be read in full here.
Plasma technique converts landfill methane to jet fuel
Trevor; there are still many landfill sites in the UK where the methane is not captured but not only that, where there is protein waste in them, they...