Point-Of-Care (POC) devices enable on-site testing by means of fast and low-cost assays that can be managed by non-skilled users. One of the main challenges in colorimetric test kits, including commercial ones, arises from the interpretation of the assay results, which often relies on the user’s capability in matching the colorimetric outcome with different reference colors. This process may lead to errors and low accuracy, practically limiting the assays only to those exhibiting strong and distinguishable colors. In addition, users’ performed readout fails in exploiting kinematic information of the colorimetric reaction that, in many cases, yields precious information about the test outcome. This talk focuses on the design of low-cost automatic systems empowered with machine-learning algorithms for the readout of colorimetric POC devices. Three use cases will be presented characterized by different constraints imposed by the kinematics of the colorimetric kit and the sensing scheme.
Edoardo Ragusa obtained the Master's degree cum laude in Electronic Engineering and a PhD in Electronic Engineering (2018), both from Genoa University, Italy. He is currently a researcher at DITEN, University of Genoa, where he teaches Digital systems Electronics and Machine Learning. He coauthored more than 40 refereed papers in international journals and conferences. He is contributing as a guest editor to Future Generation Computing Systems (Elsevier) and Electronics (MDPI). He was the technical program chair of the 6th International Conference on System-Integrated Intelligence. Intelligent, flexible and connected systems in products and production, 7-9 September Genova – Italy.
His research interests include machine learning in resource-constrained devices, convolutional neural networks, and machine learning applications.
Point-Of-Care (POC) devices enable on-site testing by means of fast and low-cost assays that can be managed by non-skilled users. One of the main challenges in colorimetric test kits, including commercial ones, arises from the interpretation of the assay results, which often relies on the user’s capability in matching the colorimetric outcome with different reference colors. This process may lead to errors and low accuracy, practically limiting the assays only to those exhibiting strong and distinguishable colors. In addition, users’ performed readout fails in exploiting kinematic information of the colorimetric reaction that, in many cases, yields precious information about the test outcome. This talk focuses on the design of low-cost automatic systems empowered with machine-learning algorithms for the readout of colorimetric POC devices. Three use cases will be presented characterized by different constraints imposed by the kinematics of the colorimetric kit and the sensing scheme.
Edoardo Ragusa obtained the Master's degree cum laude in Electronic Engineering and a PhD in Electronic Engineering (2018), both from Genoa University, Italy. He is currently a researcher at DITEN, University of Genoa, where he teaches Digital systems Electronics and Machine Learning. He coauthored more than 40 refereed papers in international journals and conferences. He is contributing as a guest editor to Future Generation Computing Systems (Elsevier) and Electronics (MDPI). He was the technical program chair of the 6th International Conference on System-Integrated Intelligence. Intelligent, flexible and connected systems in products and production, 7-9 September Genova – Italy.
His research interests include machine learning in resource-constrained devices, convolutional neural networks, and machine learning applications.
