Grant: Master's Award: Canada Graduate Scholarships
University of Toronto
$27,000.00
Heterogeneity of treatment effect in acute respiratory distress syndrome
FY 2024-25
Summary
Acute respiratory distress syndrome (ARDS) is a severe form of inflammatory acute respiratory failure. Despite representing 10% of intensive care unit (ICU) admissions and a mortality rate of approximately 40%, therapeutic options are limited. A possible explanation for the relatively high number of neutral randomized controlled trials (RCTs) in ARDS is that ARDS is a highly heterogenous clinical syndrome. Estimation of average treatment effects in RCTs fails to capture individualized treatment effects which may vary substantially based on unique clinical or biological attributes. Recent work has suggested that targeted approaches can identify variables, or groups of variables, that predict if an individual patient may derive benefit or harm from a therapeutic strategy. Our objective is to employ machine learning methods to evaluate heterogeneity of treatment effect in patients with ARDS using longitudinal clinical data collected in the Toronto Intensive Care Observational Registry (iCORE), a large registry of mechanically ventilated patients in nine University of Toronto affiliated hospitals. We will employ a causal forest machine learning algorithm incorporating all clinical and biological variables in iCORE to determine how they modulate the response to corticosteroid administration, ventilatory strategies (i.e., positive end-expiratory pressure [PEEP]), and prone positioning. This personalized approach may predict subsets of patients that could derive benefit from previously investigated treatment strategies that were discarded due to a lack of benefit in the original RCTs. There is a crucial need for this approach given the heterogeneity of ARDS, the challenges of performing RCTs in this setting, and the paucity of evidence-based interventions in this condition.
Department
Program
Master's Award: Canada Graduate Scholarships
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database
As published
Id
530302
Name
Granton David
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Heterogeneity of treatment effect in acute respiratory distress syndrome
Programtype2
Studentship Programs
Orgnameinp2
University of Toronto
Cihramount2
$27,000
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Acute Respiratory Distress Syndrome; Causal Inference; Clinical Epidemiology; Critical Care Medicine; Heterogeneity Of Treatment Effect; Machine Learning; Personalized Medicine
Abstract
Acute respiratory distress syndrome (ARDS) is a severe form of inflammatory acute respiratory failure. Despite representing 10% of intensive care unit (ICU) admissions and a mortality rate of approximately 40%, therapeutic options are limited. A possible explanation for the relatively high number of neutral randomized controlled trials (RCTs) in ARDS is that ARDS is a highly heterogenous clinical syndrome. Estimation of average treatment effects in RCTs fails to capture individualized treatment effects which may vary substantially based on unique clinical or biological attributes. Recent work has suggested that targeted approaches can identify variables, or groups of variables, that predict if an individual patient may derive benefit or harm from a therapeutic strategy. Our objective is to employ machine learning methods to evaluate heterogeneity of treatment effect in patients with ARDS using longitudinal clinical data collected in the Toronto Intensive Care Observational Registry (iCORE), a large registry of mechanically ventilated patients in nine University of Toronto affiliated hospitals. We will employ a causal forest machine learning algorithm incorporating all clinical and biological variables in iCORE to determine how they modulate the response to corticosteroid administration, ventilatory strategies (i.e., positive end-expiratory pressure [PEEP]), and prone positioning. This personalized approach may predict subsets of patients that could derive benefit from previously investigated treatment strategies that were discarded due to a lack of benefit in the original RCTs. There is a crucial need for this approach given the heterogeneity of ARDS, the challenges of performing RCTs in this setting, and the paucity of evidence-based interventions in this condition.
Competitiondate
202412
Approvedterm2
1 yr 0 mth
Pinamesdelim
Granton, David
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $27,000~ $0~