Smart Pillbox

In this project we explore the potential for using an in situ NIRS scanner placed at an elderly care centre to enable the identification of pharmaceuticals before they are administered to the patients. In such a scenario, a user can scan a pill or medicine to confirm that it is the right one to take at that moment. This is particularly important, as medication mismanagement causes substantial financial and human costs. By identifying pills, and cross-referencing them against a prescription, it is possible to reduce medication errors by ensuring that the right pill is administered at the right time.

Our device using Near-Infrared Spectroscopy to scan samples. Traditionally, NIRS has been an instrument reserved for laboratory usage, but recently affordable and smaller devices for NIRS have proliferated. Pairing this technology with a smartphone opens up a plethora of new use cases. In this project, we explore one such use case, namely medication management in a nursing home/elderly care centre.

Our work focuses on validating the performance of miniaturised NIRS for pharmaceutical identification through multiple preprocessing and classification algorithms. While there exists a large body of work dedicated to exploring the accuracy of laboratory benchtop NIRS, there is little reference literature on the just recently available miniaturised versions.

Furthermore, through user studies, we evaluate the usability of this technology when used by non-experts users. As part of this work, we have developed smartphone applications that assist users during scanning, to ensure that scanning errors and noise are minimised.

fluffy kitten
Our smart pillbox can identify unmarked pills.