Validation of electronic medical record-based phenotyping algorithms: results and lessons learned from the eMERGE network.

By K. Newton, P. Peissig, Abel N. Kho, S. Bielinski, R. Berg, Vidhu Choudhary, Melissa A. Basford, Christopher G. Chute, I. Kullo, Rongling Li and 4 more. JAMIA Journal of the American Medical Informatics Association, 2013

382 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Genetics, Health Information Management

View Paper: https://www.semanticscholar.org/paper/f70d0d8102e79d0943394ea2a5697701ce40663c

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 Genetics

In Health Information Management

Citations in the corpus

Marble: high-throughput phenotyping from electronic health records via sparse nonnegative tensor factorization (2014, background): “State of the art phenotype development, such as the eMerge Network(2), relies primarily on approaches that are heuristic, rule, and iterative based, and is a collaborative team effort between clinicians and IT experts [15, 22].”
Phenotyping Adverse Drug Reactions: Statin-Related Myotoxicity (2015, background): “Much work has been done on phenotyping disease status and raw clinical measurements (laboratory values, etc).(1) Most algorithms use ICD-9 codes alone or in concert with laboratory values, medications, and/or keywords and often require repeated measurements (i.”
Limestone: High-throughput candidate phenotype generation via tensor factorization (2014, methodology): “In this paper, we focus on diagnoses and medications due to their prominence in existing phenotype deﬁnitions [29,68].”
Electronic medical record phenotyping using the anchor and learn framework (2016, methodology): “standards, but are time consuming to build.(17) In contrast to the manually derived rules for electronic phenotyping, statistical methods, drawing on established machine learning techniques, have been used to estimate phenotypes based on inputs from the EMR.”
PheKB: a catalog and workflow for creating electronic phenotype algorithms for transportability (2016, methodology): “These studies have required accurate, sharable, and quickly adaptable phenotype algorithms implemented in varying clinical data repositories.(22) Moreover, research has shown that once created, these algorithms can be ported from one institution to another, often with comparable performance.”
Preparing Medical Imaging Data for Machine Learning (2020, background): “In general, imaging data can be labeled in a variety of ways including structured label(s), image annotations, image segmentations, and/or electronic phenotypes (1,37).”
Discriminative and Distinct Phenotyping by Constrained Tensor Factorization (2017, methodology): “After several trials, we set cardinality of the vector as 500 and window size of the sequence (i.e., the number of diagnoses or prescriptions in a sequence to consider them contextually similar) as 30.”
Extracting research-quality phenotypes from electronic health records to support precision medicine (2015, methodology): “measures an algorithm’s performance but also enhances its capability for inter-institutional sharing [36].”
Identification of type 2 diabetes subgroups through topological analysis of patient similarity (2015, methodology): “As of now, only 20 phenotypes have been validated by eMERGE (70) using iteratively refined phenotype algorithms incorporating both structured and unstructured data to achieve high PPVs to identify true cases and controls fromEMRs.”