Smart Systems

Smart AI Systems for Predictive Analytics

The research revolves around the development and implementation of intelligent AI systems designed to forecast and mitigate potential risks before they materialize. This involves leveraging advanced AI methodologies, including machine learning, deep learning, and data mining, to analyze vast datasets and identify patterns or anomalies that may indicate future problems. These systems are engineered to operate in various sectors such as healthcare, where they predict disease outbreaks or patient deterioration; manufacturing, where they anticipate equipment failures; and cybersecurity, where they detect potential security breaches. By integrating real-time data processing with sophisticated predictive models, these smart systems help organizations move from reactive to preventive strategies, fostering a more resilient and forward-thinking approach.

FEATURED WORK

SMART IoT system for obstructive sleep apnea monitoring and forecasting (joint project with spaches lab)

Members: Dr. Trung (Tim) Quoc Le, Dr. Phat Huynh, MD. Arveity Setty, Quoc Nguyen, and Quang Dang.

description

Aim 1: Develop a smart Internet of Things (IoT) platform with the analytic insights based on the system developed wearable system.

Aim 2: Collect longitudinal data and characterize the association of radiotherapy to OSA progression during cancer treatment using the developed IoT system in a sample of 20 head and neck cancer patients.

- Sensors at the edge are capable of collecting and analyze the bio-signals.

- Cross-device federated learning has been implemented at the Platform Tier.

description

The COVID-19 pandemic underscored the importance of reliable, noninvasive diagnostic tools for robust public health interventions. In this work, we fused magnetic respiratory sensing technology (MRST) with machine learning (ML) to create a diagnostic platform for real-time tracking and diagnosis of COVID-19 and other respiratory diseases. The MRST precisely captures breathing patterns through three specific breath testing protocols: normal breath, holding breath, and deep breath. We collected breath data from both COVID-19 patients and healthy subjects in Vietnam using this platform, which then served to train and validate ML models. Our evaluation encompassed multiple ML algorithms, including support vector machines and deep learning models, assessing their ability to diagnose COVID-19. Our multi-model validation methodology ensures a thorough comparison and grants the adaptability to select the most optimal model, striking a balance between diagnostic precision with model interpretability. The findings highlight the exceptional potential of our diagnostic tool in pinpointing respiratory anomalies, achieving over 90% accuracy. This innovative sensor technology can be seamlessly integrated into healthcare settings for patient monitoring, marking a significant enhancement for the healthcare infrastructure. 

Machine learning-based integrated system for intraoperative blood loss quantification in surgical sponges 

Dang NguyenS, Minh LeS, Huynh, P. K., Mohammad Al Mousa, Chinyere Charles-Okezie, Trung Q. Le, Issam El Naqa, Nguyen Quoc Khanh Le, and Aaron Muncey (2024). Machine Learning-based Integrated System for Intraoperative Blood Loss Quantification in Surgical Sponges. (Accepted for publication IEEE Journal of Biomedical and Health Informatics)


description

In the realm of surgical procedures, the accurate estimation of intraoperative blood loss remains a considerable challenge. Traditional methods, which predominantly rely on visual estimation, have significant limitations due to their inherent subjectivity and potential to underestimate actual blood loss. This can lead to delayed diagnosis of hemorrhage, inappropriate fluid resuscitation, and increased morbidity and mortality. To address this, our study introduces MDCare, a machine learning-based system specifically designed for intraoperative blood loss quantification using surgical sponges. MDCare utilizes a hardware device assembled through a combination of injection molding and blow molding, equipped with a mass sensor and a webcam for real-time monitoring. The machine learning model, trained on a large database of sponge images annotated with their sizes and levels of blood absorption, employs a Resnet-18 model to effectively recognize different sponge sizes. Additionally, MDCare uses the YOLOv4 object detection network for sponge detection, achieving an accuracy of 91.35% across nine classes of sponge sizes. The performance of the model is validated using several statistical measures, demonstrating a high level of accuracy. Clinical testing further supports the utility and accuracy of MDCare, opening up new possibilities for safer and more effective surgical procedures.

Other relevant work