Grant: Discovery Grants Program - Individual
McGill University
$20,000.00
Privacy-Preserving Analysis of Distributed Data
FY 2017-18
Summary
Data breaches and hacking can lead to the disclosure of important, sensitive, and confidential information, and can affect anyone, from celebrities to ordinary citizens. Health and clinical data breaches are especially problematic because they lead to reluctance to share existing data which may be instrumental in advancing healthcare and they may discourage individuals from participating in important clinical research. For example, disclosure of disease status for an HIV patient would be detrimental to his emotional, social and psychological well-being and could potentially demoralize him to participate in important public health research. Thus, data breaches have far-reaching negative consequences. Healthcare agencies worldwide have installed multiple safety nets to reduce data breach opportunity for hackers and to safeguard the health records of thousands of patients. One such security measure is the use of a distributed data system where provincial and/or local healthcare agencies (also called nodes) securely collect and store data in situ rather than sharing confidential health information of participants with other provinces or a central agency. However, the use of a distributed data network with multiple nodes and an analytic center presents analytical challenges. Current approach for confidential distributed data mimic meta-analyses, but this is flawed because assessments of subgroups, rare exposures, dose-response relationship, etc. are not possible. Existing Statistical Disclosure Control techniques can be used for privacy-preserving analysis; however, these methods are not well developed for distributed healthcare data, in particular where discrete and time-to-event outcomes are the norm. Suitable methods are lacking to analyze confidential distributed data. My DG research program focuses on innovative techniques to address challenges associated with analyses of such data. In particular, I propose data aggregation to preserve confidentiality of individual-level data and the use of aggregate data for statistical analysis so that combining of data will NOT be required to extract information. This novel framework will focus on non-normal disease outcomes prevalent in distributed data networks such as the Canadian Network for Observational Drug Effect Studies (CNODES). Statistical analysis strategies for aggregate data are neither well developed nor well studied. The development of a complete framework for estimation and inference, based on aggregate-level data, will include non-regular asymptotic theory, computational aspects and tools for creating aggregates, including recommendations for aggregate or group size and the development of statistical software. My research will pave a new direction for utilizing the untapped potential of rich healthcare data and will revolutionize the way information is extracted from confidential distributed data.
Program
Discovery Grants Program - Individual
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RGPIN-2017-06100
Name nom
SahaChaudhuri, Paramita
Department département
Epidemiology, Biostatistics and Occupational Health - Epidemiology, Biostatistic
Organization id
61
Institution établissement
McGill University
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2017
Competition year année de concours
2017
Award amount
20000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1205
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Mathematical sciences
Area of application fr
Sciences mathématiques
Research subject code
3005
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Biostatistics
Research subject fr
Biostatistiques
Partie
1
Nb partie
5
Application title
Privacy-Preserving Analysis of Distributed Data
Keyword
NSERC
Application summary
Data breaches and hacking can lead to the disclosure of important, sensitive, and confidential information, and can affect anyone, from celebrities to ordinary citizens. Health and clinical data breaches are especially problematic because they lead to reluctance to share existing data which may be instrumental in advancing healthcare and they may discourage individuals from participating in important clinical research. For example, disclosure of disease status for an HIV patient would be detrimental to his emotional, social and psychological well-being and could potentially demoralize him to participate in important public health research. Thus, data breaches have far-reaching negative consequences. Healthcare agencies worldwide have installed multiple safety nets to reduce data breach opportunity for hackers and to safeguard the health records of thousands of patients. One such security measure is the use of a distributed data system where provincial and/or local healthcare agencies (also called nodes) securely collect and store data in situ rather than sharing confidential health information of participants with other provinces or a central agency. However, the use of a distributed data network with multiple nodes and an analytic center presents analytical challenges. Current approach for confidential distributed data mimic meta-analyses, but this is flawed because assessments of subgroups, rare exposures, dose-response relationship, etc. are not possible. Existing Statistical Disclosure Control techniques can be used for privacy-preserving analysis; however, these methods are not well developed for distributed healthcare data, in particular where discrete and time-to-event outcomes are the norm. Suitable methods are lacking to analyze confidential distributed data. My DG research program focuses on innovative techniques to address challenges associated with analyses of such data. In particular, I propose data aggregation to preserve confidentiality of individual-level data and the use of aggregate data for statistical analysis so that combining of data will NOT be required to extract information. This novel framework will focus on non-normal disease outcomes prevalent in distributed data networks such as the Canadian Network for Observational Drug Effect Studies (CNODES). Statistical analysis strategies for aggregate data are neither well developed nor well studied. The development of a complete framework for estimation and inference, based on aggregate-level data, will include non-regular asymptotic theory, computational aspects and tools for creating aggregates, including recommendations for aggregate or group size and the development of statistical software. My research will pave a new direction for utilizing the untapped potential of rich healthcare data and will revolutionize the way information is extracted from confidential distributed data.