Feature extraction for phenotyping from semantic and knowledge resources

Ning, Wenxin; Chan, Stephanie; Beam, Andrew L.; Yu, Ming; Geva, Alon; Liao, Katherine P.; Mullen, Mary P.; Mandl, Kenneth D.; Kohane, Isaac S.; Cai, Tianxi; Yu, Sheng

doi:10.1016/j.jbi.2019.103122

Cited by 23 publications

(22 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using the PRISM features, we PRISM features, we were able to train supervised self-learning (SSL) and transfer learning (STL) classifiers that resulted in AUC ROC of 0.97, which can be compared to the specialized computational phenotyping performances between 0.94 and 0.96 in the literature. 22 , 31 , 32 …”

Section: Resultsmentioning

confidence: 99%

Generative transfer learning for measuring plausibility of EHR diagnosis records

Estiri

Vasey

Murphy

2020

Journal of the American Medical Informatics Association

View full text Add to dashboard Cite

Objective Due to a complex set of processes involved with the recording of health information in the Electronic Health Records (EHRs), the truthfulness of EHR diagnosis records is questionable. We present a computational approach to estimate the probability that a single diagnosis record in the EHR reflects the true disease. Materials and Methods Using EHR data on 18 diseases from the Mass General Brigham (MGB) Biobank, we develop generative classifiers on a small set of disease-agnostic features from EHRs that aim to represent Patients, pRoviders, and their Interactions within the healthcare SysteM (PRISM features). Results We demonstrate that PRISM features and the generative PRISM classifiers are potent for estimating disease probabilities and exhibit generalizable and transferable distributional characteristics across diseases and patient populations. The joint probabilities we learn about diseases through the PRISM features via PRISM generative models are transferable and generalizable to multiple diseases. Discussion The Generative Transfer Learning (GTL) approach with PRISM classifiers enables the scalable validation of computable phenotypes in EHRs without the need for domain-specific knowledge about specific disease processes. Conclusion Probabilities computed from the generative PRISM classifier can enhance and accelerate applied Machine Learning research and discoveries with EHR data.

show abstract

Section: Resultsmentioning

confidence: 99%

Generative transfer learning for measuring plausibility of EHR diagnosis records

Estiri

Vasey

Murphy

2020

Journal of the American Medical Informatics Association

View full text Add to dashboard Cite

show abstract

“…Phenotyping features from the EHR have been traditionally culled and curated by experts to manually construct algorithms [7], but machine learning techniques present the potential advantage of automating this process of feature selection and refinement [8][9][10][11]. Recent machine learning approaches have also combined publicly available knowledge sources with EHR data to facilitate feature curation [12,13]. Additionally, while case and control phenotyping using EHR data has also relied on a small number of expert curated cohorts, recent studies have demonstrated that ML approaches can expand upon and identify such cohorts using automated feature selection and imperfect case definitions in a high-throughput manner [14][15][16][17][18].…”

Section: Introductionmentioning

confidence: 99%

Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods

et al. 2020

View full text Add to dashboard Cite

Background Accurate identification of acute ischemic stroke (AIS) patient cohorts is essential for a wide range of clinical investigations. Automated phenotyping methods that leverage electronic health records (EHRs) represent a fundamentally new approach cohort identification without current laborious and ungeneralizable generation of phenotyping algorithms. We systematically compared and evaluated the ability of machine learning algorithms and case-control combinations to phenotype acute ischemic stroke patients using data from an EHR. Materials and methods Using structured patient data from the EHR at a tertiary-care hospital system, we built and evaluated machine learning models to identify patients with AIS based on 75 different case-control and classifier combinations. We then estimated the prevalence of AIS patients across the EHR. Finally, we externally validated the ability of the models to detect AIS patients without AIS diagnosis codes using the UK Biobank. Results Across all models, we found that the mean AUROC for detecting AIS was 0.963 ± 0.0520 and average precision score 0.790 ± 0.196 with minimal feature processing. Classifiers trained with cases with AIS diagnosis codes and controls with no cerebrovascular disease codes had the best average F1 score (0.832 ± 0.0383). In the external validation, we found that the top probabilities from a model-predicted AIS cohort were significantly enriched for AIS patients without AIS diagnosis codes (60–150 fold over expected). Conclusions Our findings support machine learning algorithms as a generalizable way to accurately identify AIS patients without using process-intensive manual feature curation. When a set of AIS patients is unavailable, diagnosis codes may be used to train classifier models.

show abstract

“…[8][9][10][11] Recent machine learning approaches have also combined publicly available knowledge sources with EHR data to facilitate feature curation. 12,13 Additionally, while case and control phenotyping using EHR data has also relied on a small number of expert curated cohorts, recent studies have demonstrated that ML approaches can expand upon and identify such cohorts using automated feature selection and imperfect case de nitions in a high-throughput manner. [14][15][16][17][18] Studies have also shown that case and control selection with diagnosis codes can signi cantly affect model performance, the hierarchical organization of structured medical data can be utilized for feature reduction and model performance improvement, and calibration is essential for understanding the clinical utility of a phenotyping model.…”

Section: Introductionmentioning

confidence: 99%

Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods

Thangaraj

Kummer

Lorberbaum

et al. 2020

Preprint

View full text Add to dashboard Cite

Background: Accurate identification of acute ischemic stroke (AIS) patient cohorts is essential for a wide range of clinical investigations. Automated phenotyping methods that leverage electronic health records (EHRs) represent a fundamentally new approach cohort identification without current laborious and ungeneralizable generation of phenotyping algorithms. We systematically compared and evaluated the ability of machine learning algorithms and case-control combinations to phenotype acute ischemic stroke patients using data from an EHR.Materials and Methods: Using structured patient data from the EHR at a tertiary-care hospital system, we built and evaluated machine learning models to identify patients with AIS based on 75 different case-control and classifier combinations. We then estimated the prevalence of AIS patients across the EHR. Finally, we externally validated the ability of the models to detect AIS patients without AIS diagnosis codes using the UK Biobank.Results: Across all models, we found that the mean AUROC for detecting AIS was 0.963±0.0520 and average precision score 0.790±0.196 with minimal feature processing. Classifiers trained with cases with AIS diagnosis codes and controls with no cerebrovascular disease codes had the best average F1 score (0.832±0.0383). In the external validation, we found that the top probabilities from a model-predicted AIS cohort were significantly enriched for AIS patients without AIS diagnosis codes (60-150 fold over expected). Conclusions: Our findings support machine learning algorithms as a generalizable way to accurately identify AIS patients without using process-intensive manual feature curation. When a set of AIS patients is unavailable, diagnosis codes may be used to train classifier models.

show abstract

Feature extraction for phenotyping from semantic and knowledge resources

Cited by 23 publications

References 57 publications

Generative transfer learning for measuring plausibility of EHR diagnosis records

Generative transfer learning for measuring plausibility of EHR diagnosis records

Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods

Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods

Contact Info

Product

Resources

About