279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction
OBJECTIVES/GOALS: The aim of this study is to analyze electronic health record (EHR) data using Mapper PLUS (MP), a new mathematical model, to classify acute myocardial infarction (MI) patients by risk of major adverse events (AE). We tested MP’s ability to define patient subgroups with distinctive...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Cambridge University Press
2023-04-01
|
| Series: | Journal of Clinical and Translational Science |
| Online Access: | https://www.cambridge.org/core/product/identifier/S2059866123003357/type/journal_article |
| _version_ | 1797840468243906560 |
|---|---|
| author | Anna Awolope Esha Datta Aditya Ballal Leighton T Izu Javier E. López |
| author_facet | Anna Awolope Esha Datta Aditya Ballal Leighton T Izu Javier E. López |
| author_sort | Anna Awolope |
| collection | DOAJ |
| description | OBJECTIVES/GOALS: The aim of this study is to analyze electronic health record (EHR) data using Mapper PLUS (MP), a new mathematical model, to classify acute myocardial infarction (MI) patients by risk of major adverse events (AE). We tested MP’s ability to define patient subgroups with distinctive risk for death, heart failure or recurrent MI after revascularization. METHODS/STUDY POPULATION: An EHR retrospective analysis of 797 MI patients and 29 variables (i.e., laboratory tests, imaging, vitals, and clinical traits) collected at the time of hospitalization was conducted. All patients received percutaneous coronary intervention and standard pharmacotherapy. MP analysis produced a multi-dimensional nodal graph of the patients based on similarities found within variables. Two algorithms, Walk Likelihood and Walk Likelihood Community Finder were applied to the graph which formed joint clusters according to spatial distance within nodes. The final output was three clusters for risk level evaluation. Risk level (low vs. high) was relative to the average risk of AEs for the entire cohort one year post MI. RESULTS/ANTICIPATED RESULTS: Of three patient subgroups, one (n= 318) had a >1 fold change for the probability of survival without AE when compared to the overall cohort and thus was defined as the low-risk group. The second group (n=304) had DISCUSSION/SIGNIFICANCE: MP stratifies patients into three groups according to predictive variables which relate to the risk for AE following an acute MI treatment. This is a new topological method for patient classification based on minimal input strictly from pre-collected EHR data. More cohort studies are needed to validate MP to classify patients for precision medicine. |
| first_indexed | 2024-04-09T16:15:35Z |
| format | Article |
| id | doaj.art-3a9bd25489f04ec196d80787431f7350 |
| institution | Directory Open Access Journal |
| issn | 2059-8661 |
| language | English |
| last_indexed | 2024-04-09T16:15:35Z |
| publishDate | 2023-04-01 |
| publisher | Cambridge University Press |
| record_format | Article |
| series | Journal of Clinical and Translational Science |
| spelling | doaj.art-3a9bd25489f04ec196d80787431f73502023-04-24T05:55:54ZengCambridge University PressJournal of Clinical and Translational Science2059-86612023-04-017838310.1017/cts.2023.335279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial InfarctionAnna Awolope0Esha Datta1Aditya Ballal2Leighton T Izu3Javier E. López4Departments of Pharmacology and Internal Medicine, University of California Davis Medical CenterDepartments of Pharmacology and Internal Medicine, University of California Davis Medical CenterDepartments of Pharmacology and Internal Medicine, University of California Davis Medical CenterDepartments of Pharmacology and Internal Medicine, University of California Davis Medical CenterDepartments of Pharmacology and Internal Medicine, University of California Davis Medical CenterOBJECTIVES/GOALS: The aim of this study is to analyze electronic health record (EHR) data using Mapper PLUS (MP), a new mathematical model, to classify acute myocardial infarction (MI) patients by risk of major adverse events (AE). We tested MP’s ability to define patient subgroups with distinctive risk for death, heart failure or recurrent MI after revascularization. METHODS/STUDY POPULATION: An EHR retrospective analysis of 797 MI patients and 29 variables (i.e., laboratory tests, imaging, vitals, and clinical traits) collected at the time of hospitalization was conducted. All patients received percutaneous coronary intervention and standard pharmacotherapy. MP analysis produced a multi-dimensional nodal graph of the patients based on similarities found within variables. Two algorithms, Walk Likelihood and Walk Likelihood Community Finder were applied to the graph which formed joint clusters according to spatial distance within nodes. The final output was three clusters for risk level evaluation. Risk level (low vs. high) was relative to the average risk of AEs for the entire cohort one year post MI. RESULTS/ANTICIPATED RESULTS: Of three patient subgroups, one (n= 318) had a >1 fold change for the probability of survival without AE when compared to the overall cohort and thus was defined as the low-risk group. The second group (n=304) had DISCUSSION/SIGNIFICANCE: MP stratifies patients into three groups according to predictive variables which relate to the risk for AE following an acute MI treatment. This is a new topological method for patient classification based on minimal input strictly from pre-collected EHR data. More cohort studies are needed to validate MP to classify patients for precision medicine.https://www.cambridge.org/core/product/identifier/S2059866123003357/type/journal_article |
| spellingShingle | Anna Awolope Esha Datta Aditya Ballal Leighton T Izu Javier E. López 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction Journal of Clinical and Translational Science |
| title | 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction |
| title_full | 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction |
| title_fullStr | 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction |
| title_full_unstemmed | 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction |
| title_short | 279 Electronic Health Record Data and Topological Data Analysis to Predict Clinical Outcomes Post Myocardial Infarction |
| title_sort | 279 electronic health record data and topological data analysis to predict clinical outcomes post myocardial infarction |
| url | https://www.cambridge.org/core/product/identifier/S2059866123003357/type/journal_article |
| work_keys_str_mv | AT annaawolope 279electronichealthrecorddataandtopologicaldataanalysistopredictclinicaloutcomespostmyocardialinfarction AT eshadatta 279electronichealthrecorddataandtopologicaldataanalysistopredictclinicaloutcomespostmyocardialinfarction AT adityaballal 279electronichealthrecorddataandtopologicaldataanalysistopredictclinicaloutcomespostmyocardialinfarction AT leightontizu 279electronichealthrecorddataandtopologicaldataanalysistopredictclinicaloutcomespostmyocardialinfarction AT javierelopez 279electronichealthrecorddataandtopologicaldataanalysistopredictclinicaloutcomespostmyocardialinfarction |