Current Projects

Panda Pain Assessment App

Undertreated post-surgical pain after discharge from hospital presents a real burden to the child, to families, and the health care system, in terms of poor outcomes and the stress of coping with the recovery at home and of potential hospital readmission. Smartphone apps have shown promise in the support of at-home management of chronic cancer pain and acute sickle cell pain in children. However, there are no apps available for the assessment and management of post-surgical pain. There is room for an efficient, accessible, and cost-effective means of providing support to caregivers to improve their management of post-operative pain. The Panda application may provide such support. This study will allow us to improve the app’s usability, and demonstrate feasibility in advance of a larger trial in which Panda will be used by caregivers and children at home.

Sepsis modeling and outcome prediction

Severe sepsis and septic shock are associated with a high burden of morbidity and mortality among critically ill children. Through improvements in supportive care, secular trends reveal a gradual decline in mortality. In collaboration with our critical care colleagues, our goal is to use a large North-American registry cohort to validate and compare the performance of existing risk score, identify new combinations of risk factors, and to develop novel outcome prediction tools.

Evaluation of data imputation strategies 

This project asks how to best deal with missing data in clinical studies. Analyses of large observational studies often encounter the issue of missing data, which are likely not missing at random. We have evaluated the performance of some pragmatic approaches when dealing with varying degrees of missingness in the clinical data needed in the calculation of two widely-accepted mortality prediction models. 

The approaches we have evaluated include using complete cases only, assuming missing data are normal, and multiple (or multivariate) imputation by chained equations (MICE). The data we have used were obtained from a large North American registry, which provides clinical markers and outcomes from curated records. The MICE approach proves to be an effective strategy with even small proportions of missing data. 

Perioperative opioid quality improvement / Pain risk prediction

There has been a dramatic increase in opioid misuse across North America. We aim to reduce the risk of surgical patients who are given an opioid prescription for their recovery from becoming long-term opioid users. We know this is a current problem with at least 6% of these patients. We plan to create a system that automatically collects and analyzes patient-centric data and provides physicians with actionable feedback on patients’ short-and long-term outcomes, including their opioid usage, opioid-sparing strategies, pain, nausea, and mobility. 

This project is focused on developing a transformative digital innovation through collaboration with industry partners. In an extension of this work, we will combine machine learning AI techniques with patient-oriented research approaches to improve the treatment of post-surgical pain in children. We propose to risk-stratify children before surgery, so that pre-habilitation and optimized analgesic combinations can be used to reduce long-term post-surgical pain.  

A Trusted, Secure and Privacy-respecting Healthcare Environment Realized for Everyone

Secure patient-controlled data access and portability of health data are limited. This industry-led project aims to empower Canadians through a trustworthy health access platform that enables them to view, share, and securely manage their health data. Our team will contribute expertise in medical device interoperability, data sharing, and technology evaluation. 

The platform we create will apply open standards and a co-operative governance model that will allow users to share confidential information through online health care services in compliance with the highest healthcare and public standards of privacy protection. Patients will control the sharing and access to the information they deem sensitive and can choose what data can be shared with whom, for what purpose, when, where, and for how long.