.svg)
The prevalence of falls among the elderly and those with limited mobility is a significant factor for personal safety and independence, one which affects the broader context of healthcare and support systems. Traditional fall detection devices, while widespread and serving a critical function, are often based on aging technologies that lack precision and sophistication.
One venerable company in elderly care recognizes the urgency of addressing this issue, and to help maximize the quality of life of their customers, has recently embarked on an innovative journey to modernize fall-detection technology.
To develop a new generation of smart fall detection devices, this company is moving to equip their devices with edge AI capabilities. Partnering with Edge Impulse, the leaders in edge AI model generation and implementation, they aim to harness the power of real-time data processing to enhance the accuracy and reliability of fall detection and to enable continually improving results over time.
The project is multifaceted, focusing on both the technological innovation and the practical application of these advancements.
“A fall is not a simple threshold, and that's why you need machine learning,” says the company’s innovation engineer. “You can't just look at a blip on X, Y and Z.”
The collaboration with Edge Impulse is key to this new initiative. The company praises the platform, stating, “We began playing with it and absolutely loved it. One of the main things is it doesn't obscure any of the details, but from a UI point of view, it appeals to new ML users. That is quite remarkable. This is the way for firmware people to do ML, but also the way for ML people to get to firmware as well.”
Additionally, Edge Impulse’s support has helped prove the viability the undertaking “The team has been brilliant. The proof of value that we did really helped to make sure that we could get a really representative application. On every front so far, the technical understanding and the support has been just amazing.”
“When the project started, I was only one person. I needed Edge Impulse. And the bill for it was cheaper than me hiring the team for me to be able to build it from scratch.”
Data-Driven Development: Leveraging vast amounts of data from users and call centers, the company can train more accurate ML algorithms, and ultimately aims for continual feedback from the next generation of devices. That would then feed back into more refined models, which would then be used to improve the device performance.
“Over time, we want to asymptotically approach the best that the hardware can do with continuous firmware updates.”
Synthetic Data: Launch a next-gen ML-based device includes innovative approaches, including exploring synthetic data generation further enhancing the model's robustness. For this, the company is looking into physics engines to model and record multitudes of falls.
Cloud integration:
With access to real world and synthetic sources, it’s a crucial need for this company to have a reliable and effortless data storage system that can support a continuous stream of new data, and then move that efficiently into Edge Impulse to refine its fall detection models. Fortunately, Edge Impulse provides seamless capabilities for integrating cloud services such as AWS S3, which allows for storing large volumes of data securely and efficiently. By leveraging AWS S3, the company can automate the uploading of new data and ensure it is readily available for Edge Impulse's advanced processing. This integration simplifies workflows, reduces latency in data analysis, and provides scalable storage options to accommodate the growing datasets necessary for refining and improving fall detection models.
The initiative is showing promising successes. Aiming to significantly reduce false positives and increase the reliability of emergency response systems, the company is, crucially, offering elderly and mobility-limited individuals a greater sense of security and independence.
The company’s edge AI project represents a significant step forward in utilizing technology to improve elderly care. By addressing the challenges of traditional fall detection systems and leveraging the power of machine learning, they are paving the way for more reliable, efficient, and innovative healthcare solutions. The collaboration with Edge Impulse demonstrates the transformative potential of edge AI in reshaping the healthcare landscape for the better.
