Taking a Breather? This Sensor Already Knows

Everything from healthcare to smart homes will benefit as human activity tracking systems become more practical for use in everyday life. These systems, which can monitor and analyze daily activities, will enable healthcare professionals to provide more personalized care and treatment plans. For example, activity tracking can help monitor chronic conditions, such as diabetes, and provide valuable insights for disease prevention and management. Additionally, smart homes will become even more convenient and efficient, as activity tracking systems can automate lighting, temperature, and security settings based on occupants' daily routines.

However, activity tracking systems are rarely seen in the wild. The reason is that they tend to be invasive in one way or another. Most existing solutions either rely on always-on cameras that compromise the user’s privacy, or they require bulky or cumbersome wearable devices that are of limited accuracy. These are not exactly the hallmarks of a technology that is going to be widely accepted by consumers.

Researchers at the Institute of Microelectronics of the Chinese Academy of Sciences have recently reported on a new type of sensor that could overcome these present issues. They have developed a novel humidity sensor that is accurate, unobtrusive, and protects the user’s privacy. It is capable of detecting even faint humidity changes in exhaled air, which can be associated with specific activities.

The newly developed sensor leverages a cutting-edge material known as porous nanoforests (NFs) to detect subtle shifts in humidity levels from human respiration. Unlike traditional sensors that struggle with stability and sensitivity, this advanced design significantly enhances accuracy. A built-in micro-heater further boosts the device’s ability to capture even the weakest respiratory changes by increasing sensitivity by 5.8 times.

Aside from being accurate, the team’s sensor is also fast. With a response time of just 2.2 seconds, it can effectively track even subtle changes in breathing patterns. This enables real-time activity classification, allowing the system to distinguish between different behaviors such as sleeping, walking, or exercising.

To make sense of the humidity data, the system employs a convolutional neural network (CNN) that processes and interprets sensor readings. By converting one-dimensional signals — such as humidity, temperature, and time — into three-dimensional maps, the CNN can classify up to nine distinct human behaviors with an impressive 96.2% accuracy.

Unlike cameras or traditional motion sensors, this approach does not rely on visual data, ensuring privacy and security for users. Since the system is integrated into a lightweight, mask-like device, it offers a discreet and comfortable way to monitor activities without interfering with daily routines. The collected data can be wirelessly transmitted to smartphones or computers, enabling real-time tracking and analysis.

This work opens the door to a wide range of applications, particularly in healthcare. Patients with respiratory conditions, such as asthma or chronic obstructive pulmonary disease, for instance, could benefit from continuous monitoring of their breathing patterns. Subtle changes in respiration could alert caregivers or medical professionals to potential health concerns before they escalate.

While the current system has demonstrated very good accuracy and practicality, future developments could further improve the technology. Enhancements in sensor miniaturization, increased battery life, and expanded behavioral classification could make this approach even more versatile.