Prediction of preterm birth in twin pregnancies using machine learning

Preterm birth (PTB) is the leading cause of death in twin pregnancies. A variety of parameters, such as cervical length, maternal medical history, demographics, and obstetric characteristics all have been shown to affect the risk of PTB. However, the relationship is not obvious. Early prediction of PTB in these pregnancies can assist physicians in identifying those patients who may benefit from preventive interventions and closer monitoring. This project aims to use machine learning to create an algorithm that predicts which twin pregnancy is at a risk of PTB.

Highly Sensitive Bacteria-Responsive Membrane Sensor Consisting of Well-aligned Core?Shell Nanofibers for In Situ Detection of Bacterial Infections

Detecting the presence of bacteria at a low concentration during a wound healing process can eliminate catastrophic results such as chronic infection and amputation. The available diagnostic techniques with high sensitivity require high-tech equipment and are expensive. Moreover, to monitor the wound with these kits/technologies, wound dressing removal is needed, which causes the second trauma. In order to overcome these shortcomings, we are fabricating Nanosheet™ Biosensor.

Development of a tele-health rehabilitation system to provide automatic assessment of patients’ performance, improve patients’ adherence and enable remote rehabilitation-(market and competitor analysis, regulatory & software design)

Poor physical recovery, especially in remote rehabilitation, is the problem that will be addressed in this project.This project is Phase I of development of Fun-exercise module. Fun-exercise system uses gamification to boost patients’ adherence to their prescribed home-exercises. To achieve this goal, Fun-exercise will use mentally stimulating and customizable games paired with a set of wearable sensors to provide feedback to ensure activities are being done correctly. In phase I, patient study and conceptual design of the Fun-exercise module will be undertaken.

Novel Artificial Intelligence (AI) driven wearable device for continuous vital sign monitoring

Wearable medical devices (SWDs) are emerging as powerful patient monitoring and data collections tools. These smart, multiplexed devices allow us to quantify dynamic biological signals in real time through highly sensitive and miniaturized biosensors. SWDs can enable monitoring at risk patients at home, diagnosing early disease progression, and reducing healthcare expenditures by means of prediction and prevention of disease.

Multi-sensor fusion for continuous vitals monitoring, sleep characterization and fall detection

The proposed research project focuses on multi-sensor such as heart rate, body temperature, oxygen saturation level, and inertial sensor data-based sleep stage classification, and tremor detection in real-time for preventive health devices. The goal of the proposed research is to build robust and energy-efficient machine learning and deep learning-based approaches that extract and analyze the significant information from the multisensor data coming from wrist-based health devices to help Parkinson's patients with tremors and an individual with sleep quality tracking.

Development of a portable and sensitive serological test for COVID-19

Luna Nanotech is in the process of developing an automated portable device for diagnosis of infectious pathogens. As part of his PhD project the intern has participated in the development of a rapid multiplex benchtop serological test for Covid-19. In this project the intern will work with Luna Nanotech scientists and engineers to adapt this benchtop serological test to be used in the diagnostic device to allow rapid automated detection of Covid-19 specific antibodies.

Spike sleep state deep learning classifier

Epilepsy is a difficult disorder to assess and even more so automatically. It is observed that in REM sleep epileptiform activity differs from other states of consciousness. REM sleep could hold the key to a better understanding of epilepsy, however robust features that work on different people and types of epilepsies are required. The purpose of using a deep learning model to classify epileptic spikes into designated sleep states can help designate those robust features.

Development of an automated and sensitive point-of-care solution for COVID-19 diagnostics

Lateral flow assays (LFA) currently used for point-of-care Covid-19 diagnostics suffer from low sensitivity and misdiagnose significant proportion of cases. However, more sensitive PCR and ELISA techniques are too time consuming or complex to be used for rapid results at point-of-care. The automated multiplex platform developed in the project will address this limitation and offer a low cost rapid 20-minute diagnostic test with the sensitivity 1000 times higher than that of LFA.

Identification, localization, and characterization of human body inflammation using thermographic imaging – phase II

The research project undertaken by the intern is to investigate the relationship between inflammation and temperature, use that knowledge to define thermal metrics and develop a standard operating procedure for 3D thermal body imaging. During the project the intern will build a phantom to evaluate and optimize the performance of a commercially available 3D thermal imager for the identification, characterization and localization of inflammation.

3D Acoustic Finite Element Modelling of the upper Airway for Obstructive Sleep Apnea Disorder Screening during Wakefulness

Obstructive sleep apnea (OSA) is an undiagnosed sleep disorder affecting up to 10% of the population. OSA patients experience morning headaches, depression and daytime sleepiness which increase the accident risk factor. Polysomnography (PSG) is the gold standard of OSA diagnosis, however, it’s expensive and time consuming. Therefore, it’s not practical to conduct PSG for patients prior to surgery requiring full anesthesia. Our research team was able to use tracheal breathing sounds for OSA diagnosis during wakefulness.