AI-powered wearable targets preventable deaths – Part 1
Avnet is working with Wearable Technologies Inc. (Wear Tech), a start-up developing a new wearable device that could drastically reduce the number of preventable deaths that occur each day.
According to the National Safety Council, there were over 3 million deaths by preventable injury in 2019 worldwide. The main preventable injuries are road injuries, falls and drownings. For older generations, the main cause of injury is a fall; for younger generations it is drowning and suffocation.
The idea for the device came from Deepak Arora, who lost his toddler daughter in a drowning accident in 2020. Society avoids talking about these incidents, but Arora realized that many similar preventable deaths happen every year. He could relate directly to how the people affected had to deal with such tragedies. With a background in technology, in part from his time working at IBM, Arora set out to develop a way to prevent them.
What are the key features of a wearable designed to prevent injury? Prevention implies foreseeing an outcome. Predicting an event would allow preventative action to be taken before it occurs. This relies on some understanding of the environmental context.
Geofencing can prevent injury
Geofencing combines a real-world location with virtualized restricted areas. A device uses this information to monitor its location within a virtually fenced area, and its proximity to hazards within that area. Geofencing relies on accurate and up-to-date location data. The fenced area can be adjusted by the user.
There are a growing number of wearable devices designed to detect events after the fact, such as falls and submersion. Some also provide location services. However, Arora’s research revealed a gap in the market for a device that can actively prevent these events, rather than merely report them.
The patent-published technology developed by Wear Tech combines geolocation data with data from biosensors (such as heart rate) and environmental sensors (air pressure, water) to understand the environmental context. The device evaluates this data to determine the wearer’s threat level.
When danger is predicted, it wirelessly connects to a cloud service to register its concerns. The alert level escalates based on the data, culminating in emergency services being notified.
Adding value with AI
Preventable injuries need to be predicted if they are to be avoided. The main goal of the Wear Tech device is to alert wearers and carers before an incident. This is fundamentally different from other products on the market. Existing products use sensor inputs to understand a situation as it is happening. Wear Tech’s device does this too, but the added value proposition is to predict the situation and prevent injuries.
To do this, the wearable needs to understand many things. One of them is relative location and nearby hazards. Another is the condition of the person wearing the device. Sensors provide the raw data for these features, but intelligence turns that data into insights.
AI running in the device is the significant differentiator. A crucial feature of the device is its use of AI to predict with high accuracy what the environmental context may be and the element of danger present.
Sensor fusion is important here. Using AI to understand what a sensor’s output means in relation to location or other environmental factors defined the Wear Tech wearable device.
Arora has leveraged open-source databases to map geolocations and identify hazards such as dangerous drops or open water. The AI algorithms developed by Wear Tech are continuously assessing the danger.
However, running AI on a constrained device is still extremely difficult. Arora, who has a background in this area, realized he would need to develop the AI algorithms in-house.
Because it uses AI, the device is always learning. It can identify a wearer’s patterns and, eventually, adjust its responses or provide reminders based on time and location. This reduces false positives.
PCIe boards offer highly dense processing power
An exploded CAD view of the Wear Tech wearable device that uses AI to help predict preventable injuries.
Inside the wearable
The three main functional elements of the device are AI, sensors and connectivity. As explained, the AI was developed entirely in-house. Arora’s experience allowed him to develop the algorithms needed to provide the predictive inferencing he was looking for.
Sensors detect the environment but also monitor the wearer. This is important because many similar devices only assume they are being worn. This means if the device isn’t being worn, it may simply assume the wearer isn’t moving. The Wear Tech device avoids this condition by using its sensors, actively confirming that it is being worn.
As well as a GPS receiver, the device uses LTE and Bluetooth 5 for communication. Using Bluetooth helps maximize battery life, while LTE is used only when necessary. This includes when the device needs to get a GPS fix (see below), but also if it is unable to connect over Bluetooth.
The wearer’s smartphone provides wide area networking. Wear Tech recognized this may not cover all use cases. The phone may be out of range, out of charge or out of coverage. Using Wi-Fi, when in range of a home network, can avoid this.
But Wi-Fi can be power-hungry, so the company is also looking at using the device’s wireless charger, which would be line-powered and stationary, as a hub or gateway. The wearable device connects to the charger through Bluetooth, while Wi-Fi in the wireless charger leverages the wide area network of the homeowner or the care home.
GPS is used to confirm location rather than continuously track the device. This also helps maximize operational time between charges. Cellular location services, coupled with dead reckoning, provide enough information for geofencing in low-risk areas. As soon as the device needs more accurate location data, it activates the GPS receiver.
How is Avnet helping Wear Tech?
Avnet has been working with Wear Tech for several years. Over that time, Avnet’s field application engineers have provided design assistance, but the real relationship is around manufacturing services.
“The core competency for me and my team is at the technology layer. We didn’t want to become a manufacturer,” Arora said. This is where Avnet’s end-to-end services are making the difference. Avnet is providing supply chain management, manufacturing, assembly, final test, and shipping.
Once the device goes into full production, customers will order the device from the Wear Tech website and everything else will be handled by Avnet.
The device is now in beta testing with customers. Arora is confident that it has the potential to make a profound impact on the world. He hopes to have the device on the market within the next year. Avnet will be following Wear Tech’s progress in this series of case studies.
AI has the potential to improve lives. AI can be used to solve some of the world's most pressing problems, such as preventable injuries. We can look forward to seeing how AI is used to make the world a safer place in the years to come.
Watch out for part 2 in this case study series, to discover the results of Wear Tech’s beta testing program.
How can Avnet help OEMs?
Avnet is primarily a component distribution company, but that isn’t where it ends. At Avnet, we understand that every electronic design and manufacturing project is different. Avnet provides products and services that can help any OEM of any size and at any point in the product lifecycle.
The relationship starts with the design cycle. Wear Tech had already engaged with a wireless module developer, but Avnet was able to help coordinate technical support between Wear Tech and the module developer thanks to the close relationships Avnet has with the semiconductor vendors involved.
The relationship with suppliers extends beyond the main module. By working with Avnet, Wear Tech gained access to the technical support provided by those suppliers through Avnet’s engineering team.
When the product reached the prototype stage and the main design choices had been made, Avnet provided proactive purchasing to ensure Wear Tech had the inventory needed to go into production.
Through its partner network, Avnet is also providing a turnkey manufacturing service. Orders will be taken by Wear Tech and passed to Avnet’s manufacturing service. The final product will be delivered by Avnet to the customer.
Avnet is also able to help Wear Tech provide aftersales support to its customers. This includes returns, repairs and warranty support.
All these services are available to OEMs around the world thanks to Avnet’s global presence and supplier relationships.