You are now in the main content area

Applying data to predict changes in intracranial pressure

A series of x-rays of a human brain, taken from multiple perspectives

Hospitals are goldmines for healthcare data, but more often than not, that data is not harvested, indexed, or examined to establish patterns or extract useful information.

To change this, computer science professor Alireza Sadeghian, the theme lead at iBEST in healthcare analytics and applications, is working with his clinical research partner, Dr. Michael Cusimano, a neurosurgeon and fellow iBEST scientist at St. Michael's Hospital in the field of traumatic brain injury.

A traumatic brain injury (TBI) can cause a person to suffer lifelong cognitive, physical, emotional, social and economic effects. Development of a reliable means of making an accurate prognosis is critical to improving our care of TBI patients.

The research partners and their team are investigating the use of advanced machine learning algorithms for analysis of the intracranial pressure (ICP) waves of TBI patients during their stay in the neurosurgery intensive care unit. They hope to determine if the machine learning approaches they've developed can decipher important aspects of the ICP waves that help to determine a patient's prognosis.

Currently, patients with a traumatic brain injury who are being monitored for intracranial brain pressure (or swelling of the brain) have a small probe inserted in order to record pressure, and that pressure is checked a handful of times over the course of the day. Professor Sadeghian predicts that the use of medical equipment could instead deliver multiple checkpoints per second, creating a more complete picture that would allow not only better health tracking but also better insight into concussions as a whole.

"The current picture they get is incomplete," said professor Sadeghian. What his team is proposing is to take the totality of the data generated by the probe in order to create an algorithm that can assess all information and look for red flags and/or patterns.

"We use deep learning techniques to predict the trend in the measurements," said professor Sadeghian. "It is a good performance indicator of how the patient is progressing."

Professor Sadeghian's research project is part of the Big Data Research, Analytics, and Information Network (BRAIN) Alliance between Ryerson University, the University of Toronto, York University, the Ontario College of Art and Design, and St. Michael's Hospital, as well as a number of industry partners. The alliance is supported by an Ontario Research Fund − Research Excellence (ORF-RE) award.

This project builds on professor Sadeghian's previous work, where he looked at changes in other biological markers, such as heart rate, pulse and blood pressure, and determined that these markers could predict a rise in intracranial pressure. Professor Sadeghian's research did show significant correlation between the markers and the ICP, making it possible to forecast elevated ICP through changes in those markers, using machine learning to accurately and quickly detect changes.