Engineers at Northwestern University in the United States have developed a new type of nanoelectronic device that can perform accurate machine learning classification tasks in the most energy-efficient way yet. The energy efficiency of this device is 100 times higher than current technology, and it can process large amounts of data in real-time and perform artificial intelligence (AI) tasks without the need to transfer data to the cloud for analysis. The research findings were published in the journal Nature Electronics on October 12th. This device has a small size, ultra-low power consumption, and no delay in receiving analysis, making it very suitable for direct integration into wearable electronic products such as smartwatches and fitness trackers for real-time data processing and almost real-time diagnosis. To test the device, researchers used it to classify information from publicly available electrocardiogram (ECG) datasets. The results show that the device can not only effectively and correctly recognize irregular heartbeat, but also determine arrhythmia subtypes from six different categories, with an accuracy rate of nearly 95%. Before machine learning tools can analyze new data, they must first accurately and reliably classify the training data into different categories. For current silicon based technology, classifying data from large datasets such as electrocardiograms requires over 100 transistors, each of which requires its own energy equipment to operate. But the nano electronic device only requires two energy devices to perform the same machine learning classification, greatly reducing power consumption. The team first trains the device to interpret electrocardiogram data, a task that typically requires trained medical staff to spend a lot of time. And this nano electronic device can accurately identify each type of arrhythmia from 10000 electrocardiogram samples. By bypassing the need to send data to the cloud, this device not only saves patients critical time but also protects their privacy. In the future, these nano electronic devices can be integrated into daily wearable devices and customized according to the health status of each user to achieve real-time applications. (New News Agency)