2 June 2021 in IEEE Access
Diabetic Retinopathy is a complication of diabetes that can cause severe vision impairment or even blindness. The condition is treatable in early stages, but it often progresses asymptomatic and undetected. Once the symptoms become noticeable, damage in the retina might already be permanent, thus many countries have implemented retinal imaging screening programmes for their diabetic population. However, existing screening programmes are not reaching all patients, and Diabetic Retinopathy stands as a leading cause of preventable blindness in the active population. We contend that the existing technological infrastructures and clinical processes may be hindering the coverage of Diabetic Retinopathy screening programmes. This paper presents EyeFundusScopeNEO, a Tele-Ophthalmology system based on a screening information system and a mobile fundus camera, which shall support opportunistic and planned screening in primary care, using cameras that are handheld, non-invasive, avoid drug-induced pupil dilation, are usable by clinicians who are not specialised in Ophthalmology, and that cost a fraction of existing table-top fundus cameras. Preliminary studies indicate the potential of the system to increase the reach of screening programmes and clinical research field trials are under preparation.
2 December 2025 in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
Deploying human activity recognition (HAR) at home is still rare because sensor signals vary wildly across houses, people, and time, essentially requiring in-situ data collection and training. Prior approaches use cameras to generate training labels for privacy-preserving sensors (LiDAR, RADAR, Thermal), but this forces sensors to detect predefined activities that cameras can see yet the sensors themselves cannot reliably distinguish. In this work, we introduce OrganicHAR, an activity discovery framework that inverts this relationship by placing sensor capabilities at the center of activity discovery. Our approach identifies naturally occurring signal patterns using privacy-preserving sensors, leverages Vision Language Models (VLMs) only during these key moments for scene understanding, and discovers discrete activity labels at granularities that these sensors can reliably detect. Our evaluation with 12 participants demonstrates OrganicHAR's effectiveness: it achieves 79% accuracy for coarse (4-5) activities using only basic ambient sensors (radar, lidar, thermal arrays), and 73% accuracy for fine-grained (8-9) activities when a wearable IMU, depth, and pose sensor are added. OrganicHAR maintains 77% accuracy on average across configurations while discovering 4-8 categories per user (15 across all users) tailored to each environment and sensor capabilities. By triggering video processing only at key moments identified by local sensors, we reduce queries to VLM by 90%, enabling practical and privacy-preserving activity recognition in natural settings.
1 June 2018 in Fusion Engineering and Design