Selected work

Representative Papers

Bridging neuroscience and AI: adaptive, culturally sensitive technologies transforming aphasia rehabilitation

Mar 22, 2026

This work addresses critical limitations in aphasia rehabilitation—namely, scarce therapeutic resources, insufficient personalization, and a lack of cultural adaptation—by integrating neuroscience and artificial intelligence with ethnographic research and neurocognitive modeling. For the first time, it incorporates linguistic diversity and cultural context into the design of AI-assisted rehabilitation tools. Employing a user-centered approach, the project developed two culturally sensitive, adaptive digital therapy prototypes that significantly enhance patient engagement. The study establishes an innovative pathway and technical framework for scalable, personalized aphasia rehabilitation that is both clinically effective and culturally grounded.

0 citationsRead paper

OTFS for Joint Radar and Communication: Algorithms, Prototypes, and Experiments

Oct 01, 2025

This paper addresses three key challenges in integrated radar-communication systems: poor multi-target resolution, low accuracy in vital sign monitoring, and difficulty distinguishing human from non-human targets. To this end, we propose an orthogonal time-frequency space (OTFS)-based joint sensing and communication architecture. Our method designs OTFS waveforms to enhance sparsity in the delay-Doppler domain, integrates a fast radar sensing algorithm with self-interference cancellation to improve multi-target separation, and combines time-frequency feature extraction with a lightweight machine learning model for high-precision respiration/heart rate estimation and human target classification. Experimental validation on an SDR platform demonstrates that the system simultaneously achieves centimeter-level ranging and millimeter-per-second-level velocity measurement for both human subjects and mobile robots. Respiratory and heart rate estimation errors are below 0.2 breath/min and 2 bpm, respectively, while human/non-human classification accuracy exceeds 96%. These results significantly extend the practical applicability of OTFS in integrated sensing and communication.

0 citationsRead paper

Recent publications

Latest Papers

Bridging neuroscience and AI: adaptive, culturally sensitive technologies transforming aphasia rehabilitation

Mar 22, 2026

This work addresses critical limitations in aphasia rehabilitation—namely, scarce therapeutic resources, insufficient personalization, and a lack of cultural adaptation—by integrating neuroscience and artificial intelligence with ethnographic research and neurocognitive modeling. For the first time, it incorporates linguistic diversity and cultural context into the design of AI-assisted rehabilitation tools. Employing a user-centered approach, the project developed two culturally sensitive, adaptive digital therapy prototypes that significantly enhance patient engagement. The study establishes an innovative pathway and technical framework for scalable, personalized aphasia rehabilitation that is both clinically effective and culturally grounded.

0 citationsRead paper

OTFS for Joint Radar and Communication: Algorithms, Prototypes, and Experiments

Oct 01, 2025

This paper addresses three key challenges in integrated radar-communication systems: poor multi-target resolution, low accuracy in vital sign monitoring, and difficulty distinguishing human from non-human targets. To this end, we propose an orthogonal time-frequency space (OTFS)-based joint sensing and communication architecture. Our method designs OTFS waveforms to enhance sparsity in the delay-Doppler domain, integrates a fast radar sensing algorithm with self-interference cancellation to improve multi-target separation, and combines time-frequency feature extraction with a lightweight machine learning model for high-precision respiration/heart rate estimation and human target classification. Experimental validation on an SDR platform demonstrates that the system simultaneously achieves centimeter-level ranging and millimeter-per-second-level velocity measurement for both human subjects and mobile robots. Respiratory and heart rate estimation errors are below 0.2 breath/min and 2 bpm, respectively, while human/non-human classification accuracy exceeds 96%. These results significantly extend the practical applicability of OTFS in integrated sensing and communication.

0 citationsRead paper