Next >

Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics

Real-World Data and Real-World Evidence: Big Data in Practice Use of EMR/EHR Database for Research and Scientific Discoveries: Procedure and Life Cycle Initiate a Project Data Queries and Data Extraction Data Cleaning Data Pre-Processing or Processing Data Preparation Data Analysis, Modeling and Prediction Result Validation Result Interpretation Publication and Dissemination Challenges and Opportunities References EHR Project Management Introduction Project and Sub-Project in EHR Research Data, Code and Product Management Team/People Management How to Form a Team: What Expertise is Needed for EHR Projects?How to Efficiently Manage a Multidisciplinary Team?Task Management Management Methods and Software Tools An Example of a Data Management Framework Folder Management File Management User Management Data Management Framework Discussion and Summary Appendix--File Submission Form Note References EHR Databases and Data Management: Data Query and Extraction Introduction EHR/EMR Database Availability and Access EHR/EMR Database Design and Structure: Database Queries Data Extraction Define Inclusion/Exclusion Criteria for Data Extraction Phenotyping: Cohort Identification Data Extraction Report Illustration Example: Subarachnoid Hemorrhage (SAH) Project EHR Database Design and Construction SAH Cohort Identification and Data Extraction Data Extraction Report Potential Data Extraction Pitfalls and Errors with Solutions References EHR Data Cleaning Introduction Review of Current Data Cleaning Methods and Tools Data Wranglers Data Cleaning Tools for Specific EHR Datasets Data Quality Assessment Common EHR Data Errors and Fixing Methods List of Common Errors in an EHR Database Demographics Table Lab Table Clinical Event Table Diagnosis and Medication Table Procedure Table Discussion Acknowledgments Notes References EHR Data Pre-Processing and Preparation Introduction Data Pre-Processing Tidy Data Principles Feature Extraction: Derived Variables Dimension Reduction Missing Data Imputation Data Preparation Define the Endpoint or Outcome Process Medical Record Timestamps Define the Encounter Time Interval Encounter Combination Define Comparison Groups Cohort Refining Leakage Detection Data Preparation for Different Analysis Purposes Data Processing/Preparation Errors and Pitfalls with Solutions Data Pre-Processing and Preparation Report Summary References Missing Data Issues in EHR Introduction and Overview Missing Data Mechanisms Methods for Incomplete EHR Data Naïve Method Imputation Using Statistical Models Machine Learning and Deep Learning Models Choice of Best Method for EHR Data Case Study Missing Condition in EHR Data Missing Imputation in EHR Datasets Evaluating the Performance of Imputation Methods and Thresholds Discussion and Conclusion References Causal Inference and Analysis for EHR Data Introduction Why Causal Inference Overview of Causal Inference Methods: Rubin Causal Model (RCM)Basic Framework in Causality: Potential Outcome Framework Propensity Scoring Brief Introduction Propensity Scoring for Binary Treatments Propensity Scoring for Multiple Treatments Propensity Scoring for Ordinal Treatments Propensity Score Estimation for Complex Data Sets Illustration Example: Subarachnoid Hemorrhage (SAH) Project Mediation Analysis Introduction to Mediation Analysis The Product Method The Difference Method Other Considerations Instrumental Variables Networks for Treatment Effect Estimation in the Presence of Unmeasured Confounders Instrumental Variables Frameworks Two-Stage Least Square Methods with Linear Models Learning Treatment Effect by Generative Adversarial Networks Introduction CGANs as a General Framework for Estimation of Individualized Treatment Effects Wasserstein GANs for Estimation of Individualized Treatment Effects MisCGANs for Estimation of Individualized Treatment Effects Optimal Treatment Selection Deconfounder in Estimation of Treatment Effects Introduction Causal Models with Latent Confounders Adversarial Learning Confounders Loss Function and Optimization for Estimating ITEs in the Presence of Confounders Targeted Maximum Likelihood Estimation Supplementary Note A Wasserstein GAN A1 Different Distances References EHR Data Exploration, Analysis and Predictions: Statistical Models and Methods Introduction Statistical Challenges for EHR Data Overview of Existing Methods Data Exploration and Visualization Statistical Models for EHR Data Contingency Tables Chi-Square Test Hypergeometric Test GLM Survival Model Mixed-Effect Models Time Series Analysis AR, MA and ARMA Model Gaussian Process Variable Selection Methods Stepwise Variable Selection Purposeful Variable Selection SIS Penalty-Based Methods Divide-and-Conquer Method Validation Results and Examples Discussions and Conclusions References Neural Network and Deep Learning Methods for EHR Data Introduction Deep Learning Methods for EHR Data Deep Learning Software Tools and Implementation Application Examples Discussion References EHR Data Analytics and Predictions: Machine Learning Methods Machine Learning Overview Machine Learning Methods Machine Learning Software Tools Application Example: SAH Project Conclusion and Recommendation References Use of EHR Data for Research: Future Future EHR Research Post-Research Practice Summary References

Next >

Neural Network and Deep Learning Methods for EHR Data
This chapter reviews the recent and emerging literature on deep learning methods in EHR data analysis. It covers a select list of methods related to structural data and discusses their advantages and disadvantages in various application scenarios and research goals. Two examples and scripts are employed...
(Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics)
Deep Learning Methods for EHR Data
DL techniques can be characterized according to their network structures. We will briefly explain a few common DL network structures and present some successful applications. The basic type of DL model is the multiple layer perceptron (MLP) (Hastie, Tibshirani, and Friedman 2001), which was popular in...
(Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics)
EHR Data Analytics and Predictions: Machine Learning Methods
Machine Learning Overview The meaning of machine learning (ML) might be different for people from different research and application fields. To simplify, here ML refers to a broad range of statistical and computational methods, theories, and tools for data exploration and analytics. In comparison...
(Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics)
Systematic Comparison of Machine Learning Methods for Identification of miRNA Species
as Disease Biomarkers Abstract. Micro RNA (miRNA) plays important roles in a variety of biological processes and can act as disease biomarkers. Thus, establishment of discovery methods to detect disease-related miRNAs is warranted. Human omics data including miRNA expression profiles have orders of magnitude...
(Bioinformatics and Biomedical Engineering)
Empirical Prediction Ability of the Machine Learning Methods
Overall discrimination results of the machine learning methods were indicated in Table 2. Calculated confusion matrices of the data sets were indicated in Table 3 (inaccurate predictions are colored with gray). “Accuracy” indicates the ratio of true positive against the entire number of samples....
(Bioinformatics and Biomedical Engineering)
Systematic Comparison of Machine Learning Methods for Identification of miRNA Species
as Disease Biomarkers Abstract. Micro RNA (miRNA) plays important roles in a variety of biological processes and can act as disease biomarkers. Thus, establishment of discovery methods to detect disease-related miRNAs is warranted. Human omics data including miRNA expression profiles have orders of magnitude...
(Bioinformatics and Biomedical Engineering)
Empirical Prediction Ability of the Machine Learning Methods
Overall discrimination results of the machine learning methods were indicated in Table 2. Calculated confusion matrices of the data sets were indicated in Table 3 (inaccurate predictions are colored with gray). “Accuracy” indicates the ratio of true positive against the entire number of samples....
(Bioinformatics and Biomedical Engineering)
MACHINE LEARNING METHODS
Machine learning approaches have been widely used for several decades for ADME/Tox and drug discovery, and they have enabled the prioritization of compounds prior to testing in vitro (Figure 17.2). The large volumes of available in vitro data have motivated the field to initially focus...
(Pharmacoeconomics: From Theory to Practice)
Choice of Best Method for EHR Data
As discussed earlier, single imputation methods, like mean or median imputation, can be used in EHR data analysis, as they are among the most straightforward ways to fill missing cells. However, these methods only exploit the information in one variable, and all are summarized in one value, independently...
(Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics)
Machine Learning Methods
In order to understand random forest, we need to start by defining the concept of a decision tree. Briefly, a decision tree is a decision-making tool that utilizes a branching, tree-like graph. Let S be the training set of n samples (xj, y:),..., (xn, yn), where Xj e X is the input instances, yt e Y...
(Statistics and Machine Learning Methods for EHR Data: From Data Extraction to Data Analytics)