ViTAL Lab

Computational BehaVior and HealTh AnaLytics (ViTAL) Lab strives to develop Patient-centered Artificial Intelligence (AI) systems that are Scalable, Accessible, and Ethical to effectively improve the Healthcare system. Our mission is to develop Edge Computing, Computer Vision, and Machine Learning systems using distributed ambient, mobile, and wearable devices to monitor patients’ conditions in hospitals or everyday life. We are also invested in deploying and testing the developed AI systems in real-world clinical and daily living environments, actively collaborating with stakeholders in healthcare.

News $\mathtt{\&}$ Talks $\mathtt{\&}$ Media

2026/05/20	📖 Preprint alert! We demonstrate AIGaitor — an on-device, end-to-end markerless motion analysis framework that runs entirely on an iPhone/iPad, without requiring cloud connectivity or external GPUs, integrating Markerless motion capture, deep learning–based gait classification, and an on-device LLM-powered interface. This work, posted on Arxiv, is led by Lauhitya Reddy with the excellent co-mentorship of Dr. Trisha Kesar from Emory Rehabilitation Medicine.
2026/05/18	📖 Preprint alert! Our first work with Shriners Children’s is posted in Arxiv, led by Lauhitya Reddy. This work demonstrates that a hierarchical deep learning model that combines a graph neural network and a transformer can quantify Rodda and Gram scores from a single-view gait video. This work has significant potential to improve consistency and scalability in caring for gait impairments in children with Cerebral Palsy (CP) in communities without expensive laboratory-based equipment.
2026/05/17	📖 Preprint alert! This work demonstrated that a hierarchical multimodal deep learning model with wearables can predict challenging behaviors in profound autism in 10 minutes advance, which provides ample time for proactive intervention. This work is posted in Arxiv and led by Yadhu Kartha. This is an excellent example of our lab’s collaboration with The Center for Discovery.