Behavior monitoring in the home

  January 31, 2023       5 min read

Chirp delivers peace of mind in a dignified way through privacy preserving technology. A key part of monitoring an aging adult is the ability to baseline behavior and automatically alert on emergencies, anomalies, and health risks. To properly baseline behavior, we need the ability to track activities of daily living (ADLs: walking, transferring, toileting, bathing, feeding, dressing), sleep, socializing, and inactivity. Tracking these behaviors in the home while balancing privacy is a challenge.

A challenge Chirp took head-on. The Chirp system is designed from the ground up with privacy in mind without compromising the benefits of behavior monitoring and automatic alerting. It starts with selecting sensors that can provide the most information while balancing privacy.

Sensors Overview

Many different sensors are used for home monitoring. Figure 1 shows a sensor’s information volume with respect to its impact on privacy. High information volume indicates the sensor can provide the necessary information for tracking ADLs and other behavior in the home.

The most common sensors used for senior monitoring are binary (on/off) sensors, such as motion sensors, contact sensors, and pressure pads. While these sensors are the most privacy-preserving, they provide the least information. For example, motion sensors can tell you there is motion in the room but cannot tell if the movement is from someone walking or turning in the bed. As a result, these sensors cannot track ADLs and other complex behaviors in the home.

On the opposite end of the spectrum are cameras. They can provide a wealth of information from gait analysis used to monitor walking to action detection for monitoring activities like transferring, feeding, and dressing. While cameras capture enough information that can be used to detect most ADLs and other complex behaviors, they come at the expense of privacy. This tradeoff is especially difficult in sensitive areas such as bathrooms and bedrooms.

Bed post sensors and sensor pads can report more than binary state information. These sensors tend to look at vital signs and sleep patterns, but they require direct contact with the person or through another object like the bed. While they provide useful information, they cannot be used to detect ADLs and other complex behaviors.

Microphones have gained significant traction within AgeTech solutions and not just for voice communication. Microphones are used for voice-based PERS (voice-activated medical alert) and to detect some ADLs based on acoustics. For example, toilet flushing could indicate toileting, and running water in the kitchen could indicate food preparation. Microphones are not as invasive as cameras. But since they listen to all sounds in the home, including speech, it is not as privacy-preserving as binary sensors such as motion sensors.

Finally, a new category of sensors in senior care solutions is radar sensors. Radar sensors, like cameras, can provide the location and movement of individuals in the home, allowing for the detection of ADLs and other complex behaviors. Unlike cameras, radars do not capture images. Furthermore, radar can detect movement through some opaque objects like shower curtains. Think of radar as high-resolution motion sensors allowing you to precisely locate the individual in the room to track them and their actions. Several types of active radar sensors are available using different frequencies, bandwidths, and algorithms.

Wi-Fi sensing is also a new form of sensing used for activity monitoring. Wi-Fi sensing has a lot of similarities with radar, but it leverages existing Wi-Fi hardware and infrastructure. Since Wi-Fi sensing leverages existing Wi-Fi communication infrastructure, the transmitted signals are designed for communication, not sensing, like radar. As a result, more complex data processing is needed for detecting movement. Wi-Fi also operates at a lower frequency than typical radar sensors, which has advantages and disadvantages. Lower frequency means motion can be detected through walls, but localizing movement to specific rooms like the bathroom becomes harder.

Sensor Combinations

Many systems today use a combination of many binary sensors for tracking basic activity in the home. As a result, they can provide information on which room has motion detected (bathroom vs. bedroom). However, they cannot detect specific ADLs and other behaviors (e.g., how sedentary, etc.).

A camera plus a microphone is also used by existing systems and can provide extensive behavior monitoring in the home. However, due to the intrusive nature of cameras, these systems are typically not used in bedrooms and bathrooms.

How do we do it?

Based on our evaluations, a radar sensor, while not as information-rich as a camera, provides privacy and enough information for Chirp’s AI system to baseline behavior. However, radar alone has a significant drawback: it is difficult for the aging adult to request help. To address this, Chirp combines a radar sensor with a microphone to detect calls for help. Furthermore, when radar is combined with acoustic information, additional ADLs (e.g., toileting, bathing) and socializing behaviors can be baselined.