Prediction Modeling Using EHR Data: Challenges, Strategies, and a Comparison of Machine Learning Approaches

By Jionglin Wu, Jason A. Roy, Walter F. Stewart. Medical Care, 2010

361 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Health Information Management

View Paper: https://www.semanticscholar.org/paper/021dc78bcb747888ad69f143516b12ca328d632b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Health Information Management

Citations in the corpus

Combining Knowledge and Data Driven Insights for Identifying Risk Factors using Electronic Health Records (2012, background): “The diagnosis date was defined as the first appearance of a HF diagnosis in the EHR [20].”
Real-world data: a brief review of the methods, applications, challenges and opportunities (2022, background): “…data-driven approaches to learn patterns, make new discoveries, assist preoperative planning, diagnostics, clinical prognostication, among others [18–27], improve predictions in selected outcomes especially if linked with administrative and claim data and usage of proper machine learning…”
Using recurrent neural network models for early detection of heart failure onset (2016, background): “9 In contrast, recurrent neural network (RNN) methods capture temporal patterns present in longitudinal data.”
Boosting Deep Learning Risk Prediction with Generative Adversarial Networks for Electronic Health Records (2017, background): “This field of research and applications, commonly mentioned as Data-Driven Healthcare (DDH) [1], has been under rapid development and attracted many researchers and institutions to utilize state-ofthe-art machine learning and statistical models on a broad set of clinical tasks which are difficult or”
Predicting diabetes-related hospitalizations based on electronic health records (2018, neutral): “Surprisingly, not until recently have EHRs been used in conjunction with advanced algorithms,[10, 11] even though they have been shown to lead to better care.”
Learning Tasks for Multitask Learning: Heterogenous Patient Populations in the ICU (2018, background): “ur experiments and results in Sections 5 and 6. 2 RELATED WORK The rapid adoption and availability of electronic health records (EHRs) has enabled new investigations into data-driven clinical support [14, 31, 44]. The broad goal of these studies is to learn from datasets of patient records in order ”
AI in Health: State of the Art, Challenges, and Future Directions (2019, background): “Conventional machine learning models for analyzing the structured information in EHRs are mostly vector based [29, 30], where patient records within a certain time window are collapsed into vectors composed of the summary statistics of the values of the features in different dimensions.”
Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records (2016, methodology): “However, predictive models and tools based on modern machine learning techniques have not been widely and reliably used in clinical decision support systems or workflows 6–9 .”
Limestone: High-throughput candidate phenotype generation via tensor factorization (2014, background): “…covering a variety of inter-related aspects of patients such as diagnoses, medication orders, and laboratory test ﬁndings [17,18]; sparsely sampled medical event sequences with different time scales across patients [19–21]; and noisy, incomplete, and inaccurate representation of patients [22,23].”
Using Electronic Health Records for Population Health Research: A Review of Methods and Applications (2015, methodology): “Better cardiac prediction has been achieved by adding variables available in the EHR that were not included in the traditional prediction models (32, 63, 93, 133).”
Preserving differential privacy in convolutional deep belief networks (2017, background): “[17, 44, 52, 65, 73], prediction of risk of readmission [73], Alzheimer’s diagnosis [46, 54], risk prediction for chronic kidney disease progression [56], physical activity prediction [57, 58, 59, 61], feature learning from fMRI data [62], diagnosis code assignment [25, 55], reconstruction of brain ”
Defining and measuring completeness of electronic health records for secondary use (2013, neutral): “Researchers may be interested in predicting, amongst other clinical phenomena, disease status and risk [36–38], readmission [39,40], or mortality [41,42].”
Using methods from the data mining and machine learning literature for disease classification and prediction: A case study examining classification of heart failure sub-types (2013, neutral): “[34], comparing the performance of logistic regression, boosting, and SVMs to predict the subsequent development of HF, found that the former two methods had comparable performance, whereas the latter method had the poorest performance.”
Clinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging. (2018, methodology): “85 Using an electronic health record-derived registry containing 536 patients, Wu et al. 86 demonstrated that logistic regression and boosting could potentially be used to predict the development of heart failure more than 6months before a formal clinical diagnosis was made, with a reported AUC of…”