Tuesday , August 3 2021

The wireless tracking system can collect health and behavior data



We live in a world of wireless signals that flow around us and give up our bodies. MIT researchers now use these signal reflections to provide scientists and carers with valuable insights into behavior and health.

The system, called Marco, transmits a low-power radio frequency (RF) signal in the middle. The signal will return to the system with certain changes if it rejects the movement of the person. The new algorithms analyze these altered reflections and link them to specific individuals.

The system then tracks the movement of each individual around the digital floor plan. Harmonizing these motion patterns with other data can provide insight into how people interact with each other and the environment.

In a paper presented at the Conference on Human Factors in Computer Systems this week, researchers describe the system and its use in the real world at six locations: two auxiliary living spaces, three dwellings inhabited by couples and one city with four inhabitants. Case studies demonstrated the ability of the system to distinguish individuals based exclusively on wireless signals – and discovered some useful behavior patterns.

In an aided dwelling, with permission from the family of the patient and caregivers, the researchers followed a patient with dementia, who was often upset for unknown reasons. Within a month, they measure the increased intensity of the patient between areas of their unit – a known sign of agitation. By appearing at an increased pace with the visitor's diary, they found that the patient was upset more during the days after family visits. This shows that Marco can provide a new passive way to monitor the functional health profiles of patients at home, researchers say.

"These are the interesting parts we discovered through data," says the first author Chen-Yu Hsu, a doctoral candidate at the Computer Science and Artificial Intelligence Laboratory (CSAIL). "We live in the sea of ​​wireless signals, and the way we move and walk changes these reflections. We have developed a system that listens to these reflections … for a better understanding of the behavior and health of people."

The research is led by Dina Kataby, Andrew and Erna Witterby, professor of electrical engineering and computer science and director of the MIT Center for Wireless Networks and Mobile Computers (Wireless @ MIT). Joining Katby and Xu on paper are graduates of MSIIL Mingmin Zhao and Guan-Hee Lee and colleague Rumen Hristov SM16.

Prediction of "tracklets" and identities

When placed in a home, Marco shoots a RF signal. When the signal jumps, it creates a type of heat map cut into vertical and horizontal "frames", indicating where people are located in a three-dimensional space. People appear as bright buds on the map. Vertical frames capture the height of the face and build, while the horizontal frames determine their general location. As individuals go, the system analyzes the RF-frames – about 30 per second – to generate short trajectories, called trays.

The convex nerve network – a machine learning model that is often used for image processing – uses these leaflets to separate reflections by certain individuals. For every individual, he feels that the system creates two "filtering masks", which are small circles around the individual. These masks essentially filter out all signals outside the circle, which locks in the trajectory and the height of the individual as they move. By combining all this information – height, construction, and movement – the network connects specific RF reflections with specific individuals.

But in order to mark the identities of those anonymous blobs, the system must first be "trained". In a few days, individuals carry low-power accelerometer sensors that can be used to mark the reflected radio signals with their respective identities. When engaged in the training, Marko first generates user manuals, as it does in practice. Then, the algorithm correlates certain acceleration functions with motion functions. When users go, for example, the acceleration oscillates with steps, but becomes a flat line when they stop. The algorithm finds the best match between the acceleration data and the track, and labels the tracking of the user's identity. Marco finds out which reflected signals correlate with specific identities.

Sensors should never be charged, and after training, individuals should not be re-used. In domain domains, Marco could mark the identities of individuals in new homes with an accuracy of 85 to 95 percent.

Achieving good (balance data) balance

The researchers hope that healthcare institutions will use Marco to passively monitor, say, how patients communicate with family and caregivers and whether patients receive medication on time. For example, in aids for living, the researchers noticed specific times that the nurse would go to the cabinet for medication in the patient's room, and then on the patient's bed. This indicates that the nurse administered the patient's medications at those particular times.

The system can also replace the questionnaires and diaries currently used by psychologists or behavioral scientists to collect family dynamics data on their subjects, daily plans or sleep patterns, among other behaviors. Those traditional recording methods may be incorrect, biased, and not well suited for long-term studies, where people may need to remember what they did for days or weeks. Some researchers have begun to equip people with a carry sensor to monitor motion and biometrics. However, elderly patients, in particular, often forget to wear or charge them. "The motivation here is to design better tools for researchers," said Hsu.

Why not just install cameras? For starters, this will require someone to see and manually record all the necessary information. Marco, on the other hand, automatically denotes patterns of behavior – such as movement, sleep and interaction – to specific areas, days and times.

Also, the video is only more interesting, adds Hsu: "Most people do not feel comfortable by being recorded all the time, especially in their home. The use of radio signals to do all this work is a good balance between getting a certain level of useful information, but does not make people feel uncomfortable. "

Katabi and her students also plan to combine Marco with the previous operation of breathing lock and heart rate from nearby radio signals. Mark will then be used to link these biometrics to the respective individuals. It can also monitor the walking speeds of people, which is a good indicator of functional health in elderly patients.

###

Additional background

APPENDIX: "Enabling identification and behavioral feelings in homes using radio reflections".
https: //dl.acm.org /quoting.cfm? id =3300778

ARCHIVE: Filling the gaps in the patient's medical records
http: // news.myth.edu /2019 /machine-learning-incomplete-patient-data-0125

ARCHIVE: a step towards personalized, automated smart homes
http: // news.myth.edu /2018 /AI-Identified-People-Closed-Smart-Homes-1017

ARCHIVE: Artificial intelligence has recalled people through the walls
http: // news.myth.edu /2018 /artificial-intelligence-senses-people-through-walls-0612

ARCHIVE: The new AI algorithm monitors sleep with radio waves
http: // news.myth.edu /2017 /new-ai-algorithm-monitors-sleep-radio waves-0807

ARCHIVE: Detect walking speeds with wireless signals
http: // news.myth.edu /2017 /dina-catabi-csail-team-develop-wireless-system-to-detect-walking-speed-0501


Source link