Fall Detection Alert

Fall Detection Alert for Elderly Accompany

What do you think most important for fall detection alert?

In order to of provide accurate data to digital health, as we all know, the fall detection alert for the elderly has always been a key topic for all of us. In the face of digital healthcare, the fall detection alert for the elderly must occupy a large market now and in the future. The core value of our products lies in the use of excellent smart hardware, including edge-cutting sensor and software algorithms. It would be easy to people take care of the elderly at home with peace of mind.

Using continuously improved internal sensors to accurately collect data, this aspect may be as everyone said, the accelerometers, gyroscopes, etc. we are familiar with, do not say much.

How to generate accurate data in fall detection alert?

Relatively speaking, the accelerometer is a relatively reliable tool for collecting data. After all, most smart machines have this module. If you consider the smart machines used by the elderly, they are not very high-end brands, and use cutting-edge accelerometers and gyroscopes. Wait, with these as the basis, it is not difficult to generate accurate fall alarm data by applying our professional algorithm experts for more than 17 years, and it is also the accumulation of technology and experience.

Furthermore, in the process of collecting data, do you need to define a placement standard? It must be held by hand, placed vertically, or placed horizontally. These influences on the magnetic field sensor or gyroscope should be relatively large. After in-depth research and development, we found that we think that from the placement position, we can eliminate large Some sensors are used.

Body and Gait for Fall Detection Alert

In addition, we have been studying the motion state of the human body, the state of the body, and the state of our steps, forming a large amount of data and accumulating a lot of experience. Almost only based on these states, we can accurately judge the motion posture, and finally we can evaluate it. A person’s exercise status, and accurate fall data, provide people with early warning reports, so that the elderly can let their families know in advance of abnormal risks, so as to make early preparations for medical testing and treatment, instead of having adverse consequences. Do testing.

Fall Algorithm Application

The fall algorithm should be, so how to use the fall algorithm? Some people say that a mature decision tree algorithm can be used. Today, machine learning is so popular, and it can also be used in this project. Decision tree c4.5 algorithm can be considered. Or the recurrent neural network, if we collect a small sample that is often 1 second, the sampling frequency is 50Hz, and one data is 50*3, which is 150 floating-point numbers, which is much simpler than complex image processing. From the perspective of operation speed That said, there is no problem implementing real-time motion detection.

It is great to transplant the trained model to the cloud platform and mobile application, so this algorithm for fall alarm is also our experience. It has also been recognized by many customers.

In fact, in the selection of sensors and algorithms, we have already got good expectations. The key is, how to collect the data set? We have analyzed various fall postures earlier. Yes, the data of falls are very simple, but the analysis is very complicated.

Collecting data sets, normal exercise data sets are easy to handle. I bring an app that can collect data in real time, and I take it with me when I walk, eat, sleep, and exercise. As long as I don’t fall down, all the data can be used as non-fall data. Come to do classification statistics, and more work is in our algorithm. If you need to know more, please contact us.

Collection and classification for fall detection algorithm

But what about falling datasets? We all use tens of thousands of data collections to form, we can’t let nursing homes or the elderly do the test. Then it is not unreasonable to use the fall data of young people as a substitute. Although it is not the data of falls of the elderly, it is acceptable.

But these tests were conducted on simulated elderly people. In addition, we also actually tested various categories of elderly people, different ages, different regions, different heights, different genders, different weights, and different physical health status. These tests are the basis of our cloud data. With these, we can do artificial intelligence algorithms. In order to have accurate fall data,

For the collection of data sets, we establish motion pose models of the elderly at various ages and conduct verification tests in various situations. Finally, we can provide our products to the market, so that everyone can safely apply them to your own brand of digital health.

If you are interested, you can contact us, or visit our OEM and ODM hiedesign website to learn more about our group’s comprehensive resources in electronic solutions.