Medication trackers call for data fusion secret recipe
Taking various shapes such as a wraparound rubber band for bottled or pressurized can medication (inhalers or pen-shaped insulin injectors), or delivered as a clip-on or stick-on device for blisters and pill boxes, Amiko’s medication activity tracker logs medication intakes and links the data to a smartphone app for scheduling, history-keeping and reminder alerts.
The data can also be securely shared with close relatives, care-takers or physicians. Nothing new if you consider the plethora of activity trackers and apps around that could do a similar job (see Internet cookies move to the physical world).
“The hardware architecture is mostly similar to other equivalent products used for fitness and body tracking using MEMS data” admits Alessandro Fato, Lead Firmware Engineer at Amiko, “but where we make a real difference is the level of reliability we provide when making assumptions on medication intakes”.
“In a fitness application, it doesn’t really matter if you miss a heart-beat or a few burnt calories, it is not life-threatening”, explains Fato, “but regarding medication, we want to be able to tell you for sure if you took or not the medication”.
Working closely with STmicro, the company has developed and patented special data fusion algorithms that it claims are much more reliable than competing offerings while remaining very power-efficient.
“We are performing high frequency data tracking and filtering, but at a very low cost and very low power,” emphasized Fato, “and this is the main challenge for a battery-powered solution as these requirements are typically in conflict”. The trackers are expected to run continuously from four to six months on a large coin cell battery.
The company has compiled a database of empirical sensor data (opening blisters, using insulin pen-injectors, inhalers etc..) to fingerprint and match the related high acceleration patterns of such medication delivery mechanisms and distinguish them from background noise such as a medication package being carried around and shaken in a handbag or subject to automotive vibrations.
Amiko is currently working on a generic device that will be able to be parameterised for different styles of medication packages, which in principle would increase the tracking reliability by reducing the number of possible detection patterns (cracking a blister open or pressing the cap of an inhaler).
The startup wants to develop modular and extensible configurable firmware to adapt its solution to the different types of medication packages, understanding how the different materials and geometries can impact the tracker’s response.
It is difficult to compare data fusion algorithms and power efficiency since there are no firm benchmarks, it is all about what compromises you are ready to make.
But Fato is confident that Amiko’ secret recipe will find many other niche applications where a specific activity signal must be identified with the utmost reliability. In the future, Fato hopes to have Amiko’s deterministic firmware modules formally certified for high reliability applications.
Another point Fato highlighted is that contrarily to dedicated pill-dispenser boxes, Amiko’s tracker device doesn’t require the user or the care-taker to move pills or liquids around. The medication stays in its original package, which makes it easier for medical certification since the drugs stay within the same controlled atmosphere as they are delivered by the pharmaceutical companies. The trackers also include climate-sensitive data (assume temperature and humidity) and can be programmed to send expiry date alerts.
Visit Amiko ‘s official page at www.amiko.io
Related articles:
Internet cookies move to the physical world
MEMS’ new battleground: Hardware-agnostic sensor fusion?
IoT: sensor fusion or confusion?
IoT cybersecurity: is EDA ready to deliver?
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
