UNB Researcher’s Innovation Could Forecast Accidents Before They Happen
Research from a computer science PhD candidate at the University Of New Brunswick (UNB) could help wearable health monitors forecast possible accidents before they even happen.
Alireza Manashty recently published a paper that looks at a model that could store years of health records for a person in a short sequence so that small wearable devices could process the information using state-of-the-art deep learning models.
“This novel model is called Life Model for Time Series (LMts) which enables forecasting events, versus just detecting, in order to prevent severe accidents, like elderly falls, heart attacks or seizures,” says Manashty.
This means that wearable devices can store and process more data from a longer period of time in order to provide the wearer with more accurate insights and predictions based on their personal profile.
“For example, if a person has suffered from a heart problem years ago, a higher BPM [beats per minute] could do more harm than good compared to a similar individual with no such history,” says Manashty. “Moreover, their devices could now process all this information using deep learning models, whereas before, it generally required high-end computers.”
When he is not working on his research, Manashty works as a data science consultant for government, the Saint John Regional Hospital, and the university to help them leverage data using machine learning, deep learning, and artificial intelligence. He says doing his research at UNB has given him a new approach to his work along with strong support.
“I am honored to work with unique and experienced researchers and individuals at UNB, especially my supervisor Professor Janet Light, who contributed to elderly fall research and supported this research and Professor Chris Baker with his research in semantic health and ongoing support for research in the computer science department,” says Manashty. “There’s a diversified research environment that involves researchers from engineering and epidemiology which enables any researcher to think beyond their field of study.”
Last month, Manashty was recognized by the New Brunswick Health Research Foundation (NBHRF) as one of its “Rising Stars.” He says the organization provides opportunities for researchers in the province the New Brunswick.
“NBHRF’s annual conference is a great venue for health researchers to meet and collaborate,” he says. “Also recognizing top researchers encourages everyone in the province to keep up the pace with the top performers.”
Next, Manashty says he is now working on turning his research into a functioning product.
“The next step in my research is to create a working prototype of a wearable sensor to enable activity and human fall forecasting in order to predict and prevent emergency accidents,” he says.