A wearable, continuous electromagnetic sensing system provides 96% glucose estimation accuracy when tested on healthy individuals.

This paper presents a new compact lightweight radio frequency (RF) energy harvesting system. The system relies on a dual-tapered transmission line-based matching network that stretches the rectification capability of an integrated Schottky diode. This topology is demonstrated on a 2.4 GHz rigid harvesting system with a resulting power conversion efficiency that reaches up to 58% over 0 dBm input RF power. Its performance is compared with a reference rectifier that relies on a typical open circuit shunt-stub matching network. The rectifier along with a miniaturized monopole antenna are then tailored for a flexible substrate with a resulting efficiency around 50% at 0 dBm input power. The rectifier exhibits an almost flat efficiency over the 2.3-2.5 GHz frequency span despite a wide load variations. The system is characterized in multiple bent configurations featuring high and stable performance. It is demonstrated that for different bent states, the proposed flexible harvester does not display large variations in harvested power. Thus, the presented rectenna demonstrates the remarkable combination of compactness, flexibility and stability. Equipped with these features, such rectifier can be plugged into a variety of sensors, even on wearable surfaces; which makes it ideal for Internet of Things (IoT) applications.

In this paper, a novel, highly accurate, non-invasive glucose-monitoring sensor that is based on a flexible broadband reject filter is presented. The filter topology comprises a tapered feed line at a top layer that excites four modified log-periodic open loop resonators on the bottom layer, achieving a broadband reject response. Size reduction techniques are applied on the embedded resonators that are optimized to exhibit an enhanced sensitivity to track the variations of the glucose level across a frequency span from 1.25 GHz to 2.65 GHz. The proposed flexible filter is tested pre-clinically and clinically, where a high correlation between its scattering parameters and the variations in glucose levels is attained. Regression models are also developed using experimental data obtained from healthy patients that are subjected to glucose tolerance tests. Results demonstrate less than 4% mean absolute relative difference between the reference and estimated glucose levels, and the predicted glucose levels lie 100% within the clinically acceptable zones as shown by the Clarke Error Grid analysis.

The work presented herein proposes an alternative skin cancer screening method that delivers non-invasive diagnosis and monitoring of skin lesions by leveraging electromagnetic waves with radio frequency technology and circuits. The proposed handheld device, named SkanMD, comprises a sensitive electromagnetic sensor, customized radio frequency wave analyzer circuits, and machine learning algorithms. The device is used in clinical studies that are performed on a total of 46 individuals that are composed of 18 patients with pre-diagnosed skin cancer, 10 individuals with benign nevi, 7 patients with arbitrary diseases, and 11 healthy individuals. These studies included the measurement of the reflection coefficient, S11, on multiple skin regions and recording the obtained complex values to build a Support Vector Machine (SVM)-based classification model. Due to the lesion-optimized sensor and the unified cross-patient classifier, our results differentiate between cancerous and non-cancerous skin lesions with a sensitivity that exceeds 92% and a specificity that exceeds 81.4%. These reported results are based on a limited population size study. They also demonstrate that SkanMD is a promising solution that could augment conventional diagnosis methods to greatly improve patient comfort and enable instantaneous and accurate diagnosis.