Mutlu Mete

Anuradha Chandrasekaran, Dimitrios Zikos, Mutlu Mete et al.

Clinical AI adoption is hindered by the black-box/grey-box nature of high-performing models, which lack the ontological grounding and narrative transparency required for professional-level explainability. We present NEURON, a neuro-symbolic system designed to enhance both predictive reliability and clinical interpretability. NEURON integrates SNOMED CT ontology-informed structural representations with machine learning models to bridge the gap between raw data and medical nomenclature. To facilitate human-aligned interaction, the system utilizes a Retrieval-Augmented Generation (RAG) grounded LLM layer to synthesize SHAP feature attributions and patient-specific clinical notes into coherent, natural-language explanations. Validated on the MIMIC-IV dataset for Acute Heart Failure mortality prediction, NEURON improved the AUC from 0.74-0.77 to 0.84-0.88 and significantly outperformed raw SHAP visualizations in human-aligned metrics (0.85 vs. 0.50). Our results demonstrate that NEURON offers a robust, scalable engineering solution for deploying trustworthy, human-centered connected health applications.

Kathiravan Natarajan, Truong-Huy D. Nguyen, Mutlu Mete

Drones are conventionally controlled using joysticks, remote controllers, mobile applications, and embedded computers. A few significant issues with these approaches are that drone control is limited by the range of electromagnetic radiation and susceptible to interference noise. In this study we propose the use of hand gestures as a method to control drones. We investigate the use of computer vision methods to develop an intuitive way of agent-less communication between a drone and its operator. Computer vision-based methods rely on the ability of a drone's camera to capture surrounding images and use pattern recognition to translate images to meaningful and/or actionable information. The proposed framework involves a few key parts toward an ultimate action to be taken. They are: image segregation from the video streams of front camera, creating a robust and reliable image recognition based on segregated images, and finally conversion of classified gestures into actionable drone movement, such as takeoff, landing, hovering and so forth. A set of five gestures are studied in this work. Haar feature-based AdaBoost classifier is employed for gesture recognition. We also envisage safety of the operator and drone's action calculating the distance based on computer vision for this task. A series of experiments are conducted to measure gesture recognition accuracies considering the major scene variabilities, illumination, background, and distance. Classification accuracies show that well-lit, clear background, and within 3 ft gestures are recognized correctly over 90%. Limitations of current framework and feasible solutions for better gesture recognition are discussed, too. The software library we developed, and hand gesture data sets are open-sourced at project website.