Home
|
Inbox
|
Search
|
View Abstract
Genomics Analysis of Gene Expression Profiles Demonstrates a Distinct ARDS Signature
Description
.abstract img { width:300px !important; height:auto; display:block; text-align:center; margin-top:10px } .abstract { overflow-x:scroll } .abstract table { width:100%; display:block; border:hidden; border-collapse: collapse; margin-top:10px } .abstract td, th { border-top: 1px solid #ddd; padding: 4px 8px; } .abstract tbody tr:nth-child(even) td { background-color: #efefef; } .abstract a { overflow-wrap: break-word; word-wrap: break-word; }
A7533 - Genomics Analysis of Gene Expression Profiles Demonstrates a Distinct ARDS Signature
Author Block: J. A. Howrylak1, V. Walter2, E. Wasserman2, M. Moll3, T. Dolinay4, L. Schultz1, Z. Ma1, A. M. Choi5, R. M. Baron6, N. J. Thomas7, H. R. Wong8, J. R. Broach9, V. M. Chinchilli10; 1Medicine, Penn State University, Hershey, PA, United States, 2Department of Public Health Sciences, Penn State University, Hershey, PA, United States, 3Department of Pulmonary and Critical Care Medicine, Brigham and Woman's Hospital, Boston, MA, United States, 4Department of Pulmonary and Critical Care Medicine, Hospital of the University of Pennsylvania, Wynnewood, PA, United States, 5Department of Medicine, Weill Cornell Medical College, New York, NY, United States, 6Department of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, Brookline, MA, United States, 7Department of Pediatrics, Penn State University, Hershey, PA, United States, 8Department of Pediatris, Childrens Hosp Med Ctr, Cincinnati, OH, United States, 9Biochemistry and Molecular Biology, Penn State University, Hershey, PA, United States, 10Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, United States.
Rationale: The acute respiratory distress syndrome (ARDS) is a complication of critical illness and accurate diagnosis remains challenging. Lack of objective molecular biomarkers for ARDS hampers the selection of study subjects for large-scale clinical trials as well as the development of novel therapeutic options.
Objective: To use gene expression profiles from critically ill patients to identify a molecular signature for ARDS.
Methods: We pooled gene expression profiles from five publicly available datasets from critically ill adult and pediatric patients with and without ARDS and applied machine learning techniques to a discovery cohort of subjects to develop a predictive model of ARDS that we subsequently validated in an independent cohort of subjects.
Measurements and Main Results: We used multiple machine learning methods to develop a predictive model for ARDS. We constructed random forest models using 1000 iterations of a model building and testing approach in a training cohort (75%, n = 318), and identified genes with high importance measures. We next built a final random forest using expression data from training cohort and the 100 top genes from the iterated analysis. When applied to a separate validation cohort (25%, n = 105), the sensitivity and specificity of this model at identifying ARDS patients in the validation cohort were 73.7% and 85.1%, respectively.
Conclusions: This study demonstrates the potential of gene expression profiling to improve ARDS diagnosis in the ICU, and has implications for mechanistic studies, medical decision-making and the development of future clinical trials. Prospective validation is necessary to confirm these findings.
Author Block: J. A. Howrylak1, V. Walter2, E. Wasserman2, M. Moll3, T. Dolinay4, L. Schultz1, Z. Ma1, A. M. Choi5, R. M. Baron6, N. J. Thomas7, H. R. Wong8, J. R. Broach9, V. M. Chinchilli10; 1Medicine, Penn State University, Hershey, PA, United States, 2Department of Public Health Sciences, Penn State University, Hershey, PA, United States, 3Department of Pulmonary and Critical Care Medicine, Brigham and Woman's Hospital, Boston, MA, United States, 4Department of Pulmonary and Critical Care Medicine, Hospital of the University of Pennsylvania, Wynnewood, PA, United States, 5Department of Medicine, Weill Cornell Medical College, New York, NY, United States, 6Department of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, Brookline, MA, United States, 7Department of Pediatrics, Penn State University, Hershey, PA, United States, 8Department of Pediatris, Childrens Hosp Med Ctr, Cincinnati, OH, United States, 9Biochemistry and Molecular Biology, Penn State University, Hershey, PA, United States, 10Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, United States.
Rationale: The acute respiratory distress syndrome (ARDS) is a complication of critical illness and accurate diagnosis remains challenging. Lack of objective molecular biomarkers for ARDS hampers the selection of study subjects for large-scale clinical trials as well as the development of novel therapeutic options.
Objective: To use gene expression profiles from critically ill patients to identify a molecular signature for ARDS.
Methods: We pooled gene expression profiles from five publicly available datasets from critically ill adult and pediatric patients with and without ARDS and applied machine learning techniques to a discovery cohort of subjects to develop a predictive model of ARDS that we subsequently validated in an independent cohort of subjects.
Measurements and Main Results: We used multiple machine learning methods to develop a predictive model for ARDS. We constructed random forest models using 1000 iterations of a model building and testing approach in a training cohort (75%, n = 318), and identified genes with high importance measures. We next built a final random forest using expression data from training cohort and the 100 top genes from the iterated analysis. When applied to a separate validation cohort (25%, n = 105), the sensitivity and specificity of this model at identifying ARDS patients in the validation cohort were 73.7% and 85.1%, respectively.
Conclusions: This study demonstrates the potential of gene expression profiling to improve ARDS diagnosis in the ICU, and has implications for mechanistic studies, medical decision-making and the development of future clinical trials. Prospective validation is necessary to confirm these findings.
|
Home
|
Inbox
|
Search
|