Consumables are among the most challenging items to track, with manual reporting often inaccurate, inefficient, or both. Poor estimations can in turn lead to under- or over-ordering of inventory, and as such the ability for brand owners and retailers to remotely and accurately monitor and track customer usage of consumables can provide enormous business benefits. Product usage and consumption tracking also extends beyond inventory management to a whole host of potential applications including medication adherence, abuse prevention, as well as market research based on clinical trials.
To that end, Candibell, a Boston, MA-based commerce automation platform developer, identified an opportunity to develop a “toolbox for remote usage monitoring of physical things”. The resulting Candibell sensing modules, which come in reusable as well as ‘one-time use only’ models, were designed to fill this sizeable gap in the market.
Combining IoT, sensory and pattern recognition technologies with mathematical modelling and low power wireless connectivity, Candibell is a unique, scalable solution for ‘real time’ remote analysis of comprehensive usage data for pharmaceutical and consumable products. In addition, brands and vendors have access to the complete consumption history of an individual customer, patient or user, helping to inform future decision making. But Candibell is more than a consumption tracking solution alone - it also provides a platform for integrating automated replenishment and other related services.
“One can think of Candibell as a kind of ‘fitness tracker’ for a physical product,” says Eric Huang, President and CEO, Candibell. “In the digital world, for example, we have common services for website traffic analysis and mobile app screen time tracking etc. Similarly, Candibell provides usage tracking and event-based automation for consumable products, but in the physical world.”
Valid consumption events
Candibell sensing modules integrate an accelerometer and gyroscope that continuously monitor the movement of a product package to detect any potential “consumption event” using Candibell’s proprietary algorithms. The motion data is recorded and relayed using Nordic SoC-enabled Bluetooth LE connectivity to a dedicated ‘Candibell Hub’ gateway or smartphone, and subsequently sent to the company’s Cloud platform. There are two versions of Candibell Hub: One using Wi-Fi to connect to the Cloud; and the other using LTE-M low power cellular technology.
At this point the platform determines whether or not the sequence of movement—for example if a pill container is inverted or a blister pack removed from its external packaging—appears to be a valid consumption event. The verified product usage data and consumption history information is then accessible in ‘real time’ via a web-based dashboard or application programming interface (API). At the same time, vendors and brands can subscribe to a set of event data that enables them to implement additional automated e-commerce services like product replenishment.
But that’s not the limit of Candibell’s capabilities. The solution also provides an iOS and Android SDK allowing vendors to integrate the data into their own smartphone apps to provide individuals, or their caregivers, visibility of consumption data. For example, patients living independently with chronic diseases could use Candibell to provide caregivers and family members the ability to remotely monitor their wellbeing, such as consumption of food, medication and personal hygiene products.
As the sensing modules use motion algorithms to detect the consumption events, the same module can be used for multiple package designs such as bottles of different sizes, blister packs, boxes and bags, as well as for products of different forms including liquid, lotion, pharmaceutical solids and powders.
“Our small, low cost and power efficient sensing modules can be built organically into the product’s package. Alternatively, they can be attached to products at retail level,” Huang explains.
Motion signal processing
Thanks to the Nordic SoC’s Arm Cortex M4 processor, which supports the application’s complex Floating Point and Digital Signal Processing computations, the SoC is able to comfortably run the Candibell device’s motion signal processing before relaying the data to the gateway via Bluetooth LE.
“The nRF52832 SoC was the ideal candidate for our sensing module because as well as being low cost, it is packed with a lot of computational power, a large memory capacity and extremely low power consumption in standby mode,” Huang says. “All of these factors helped us to design a product that is cost effective, small and battery efficient, making the product more competitive.”
Moreover, the company is quick to acknowledge the wider role played by Nordic’s experience and expertise in the successful development and launch of the Candibell sensors.
“Nordic’s comprehensive documentation, code examples, SDK and powerful Nordic Thingy:52 development platform made fast prototyping extremely easy, greatly simplifying our job,” Huang says. “All we needed to focus on was our own business logic.”