Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Master's Award: Canada Graduate Scholarships

University of Calgary

$17,500.00

Enabling the Analysis of Patient-Level Data Across Regional Boundaries for Research Use: An Exploration of the Utility of the Likelihood Function

FY 2023-24

Summary
In Canada, the disclosure of personal health information and de-identified data are subject to specific provincial and territorial legislation. Legislative and policy limitations exist for the disclosure, transfer, and sharing of patient level data across regional boundaries for research use. With very few exceptions, individual-level health data is not allowed to cross provincial boundaries. Currently, meta-analysis techniques are used to pool province-specific results to enable national-level results while maintaining compliance with the legislative restrictions. However, there are many limitations to this approach. Therefore, a more effective methodology is needed to enable the analysis of health information without sharing data, thus allowing sensitive health data to remain private and secure. One possibility includes an adaptation of federated learning through a statistical characteristic known as the likelihood function. The likelihood function carries all the information contained in the sample data about that model, while being sufficiently aggregated to ensure no personal information is able to “leak through.” According to likelihood theory, the likelihood functions for the various jurisdictional models can be combined to yield an overall likelihood function that is mathematically identical to what would be available if the data were all combined for the analysis. For my master’s thesis, I propose to explore the utility of the likelihood function to enable legislatively compliant provincial models, while maintaining the statistical advantages of a fully pooled analysis. I will compare the performance of this method to the existing meta-analysis approach.
Program
Master's Award: Canada Graduate Scholarships
Location
Alberta, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database

As published

Id
506074
Name
Harmon Megan
Region
Alberta
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Enabling the Analysis of Patient-Level Data Across Regional Boundaries for Research Use: An Exploration of the Utility of the Likelihood Function
Programtype2
Studentship Programs
Orgnameinp2
University of Calgary
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Biostatistics; Chronic Kidney Disease; Electronic Medical Records; Likelihood Function; Methodology; Primary Care; Privacy; Provincial And Territorial Legislation; Regional Boundaries; Security
Abstract
In Canada, the disclosure of personal health information and de-identified data are subject to specific provincial and territorial legislation. Legislative and policy limitations exist for the disclosure, transfer, and sharing of patient level data across regional boundaries for research use. With very few exceptions, individual-level health data is not allowed to cross provincial boundaries. Currently, meta-analysis techniques are used to pool province-specific results to enable national-level results while maintaining compliance with the legislative restrictions. However, there are many limitations to this approach. Therefore, a more effective methodology is needed to enable the analysis of health information without sharing data, thus allowing sensitive health data to remain private and secure. One possibility includes an adaptation of federated learning through a statistical characteristic known as the likelihood function. The likelihood function carries all the information contained in the sample data about that model, while being sufficiently aggregated to ensure no personal information is able to “leak through.” According to likelihood theory, the likelihood functions for the various jurisdictional models can be combined to yield an overall likelihood function that is mathematically identical to what would be available if the data were all combined for the analysis. For my master’s thesis, I propose to explore the utility of the likelihood function to enable legislatively compliant provincial models, while maintaining the statistical advantages of a fully pooled analysis. I will compare the performance of this method to the existing meta-analysis approach.
Competitiondate
202312
Approvedterm2
1 yr 0 mth
Pinamesdelim
Harmon, Megan
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~