Phe2vec: Automated disease phenotyping based on unsupervised embeddings from electronic health records.
Ontology highlight
ABSTRACT:
SUBMITTER: De Freitas JK
PROVIDER: S-EPMC8441576 | biostudies-literature | 2021 Sep
REPOSITORIES: biostudies-literature
Publications
Phe2vec: Automated disease phenotyping based on unsupervised embeddings from electronic health records.
De Freitas Jessica K JK Johnson Kipp W KW Golden Eddye E Nadkarni Girish N GN Dudley Joel T JT Bottinger Erwin P EP Glicksberg Benjamin S BS Miotto Riccardo R
Patterns (New York, N.Y.) 20210902 9
Robust phenotyping of patients from electronic health records (EHRs) at scale is a challenge in clinical informatics. Here, we introduce Phe2vec, an automated framework for disease phenotyping from EHRs based on unsupervised learning and assess its effectiveness against standard rule-based algorithms from Phenotype KnowledgeBase (PheKB). Phe2vec is based on pre-computing embeddings of medical concepts and patients' clinical history. Disease phenotypes are then derived from a seed concept and its ...[more]
Similar Datasets
Project description:ObjectiveUnstructured data present in electronic health records (EHR) are a rich source of medical information; however, their abstraction is labor intensive. Automated EHR phenotyping (AEP) can reduce the need for manual chart review. We present an AEP model that is designed to automatically identify patients diagnosed with epilepsy.MethodsThe ground truth for model training and evaluation was captured from a combination of structured questionnaires filled out by physicians for a subset of patients and manual chart review using customized software. Modeling features included indicators of the presence of keywords and phrases in unstructured clinical notes, prescriptions for antiseizure medications (ASMs), International Classification of Diseases (ICD) codes for seizures and epilepsy, number of ASMs and epilepsy-related ICD codes, age, and sex. Data were randomly divided into training (70%) and hold-out testing (30%) sets, with distinct patients in each set. We trained regularized logistic regression and an extreme gradient boosting models. Model performance was measured using area under the receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), with 95% confidence intervals (CI) estimated via bootstrapping.ResultsOur study cohort included 3903 adults drawn from outpatient departments of nine hospitals between February 2015 and June 2022 (mean age = 47 ± 18 years, 57% women, 82% White, 84% non-Hispanic, 70% with epilepsy). The final models included 285 features, including 246 keywords and phrases captured from 8415 encounters. Both models achieved AUROC and AUPRC of 1 (95% CI = .99-1.00) in the hold-out testing set.SignificanceA machine learning-based AEP approach accurately identifies patients with epilepsy from notes, ICD codes, and ASMs. This model can enable large-scale epilepsy research using EHR databases.
| S-EPMC10239346 | biostudies-literature
Project description:Bipolar disorder (BD) is a heritable mood disorder characterized by episodes of mania and depression. Although genomewide association studies (GWAS) have successfully identified genetic loci contributing to BD risk, sample size has become a rate-limiting obstacle to genetic discovery. Electronic health records (EHRs) represent a vast but relatively untapped resource for high-throughput phenotyping. As part of the International Cohort Collection for Bipolar Disorder (ICCBD), we previously validated automated EHR-based phenotyping algorithms for BD against in-person diagnostic interviews (Castro et al. Am J Psychiatry 172:363-372, 2015). Here, we establish the genetic validity of these phenotypes by determining their genetic correlation with traditionally ascertained samples. Case and control algorithms were derived from structured and narrative text in the Partners Healthcare system comprising more than 4.6 million patients over 20 years. Genomewide genotype data for 3330 BD cases and 3952 controls of European ancestry were used to estimate SNP-based heritability (h2g) and genetic correlation (rg) between EHR-based phenotype definitions and traditionally ascertained BD cases in GWAS by the ICCBD and Psychiatric Genomics Consortium (PGC) using LD score regression. We evaluated BD cases identified using 4 EHR-based algorithms: an NLP-based algorithm (95-NLP) and three rule-based algorithms using codified EHR with decreasing levels of stringency-"coded-strict", "coded-broad", and "coded-broad based on a single clinical encounter" (coded-broad-SV). The analytic sample comprised 862 95-NLP, 1968 coded-strict, 2581 coded-broad, 408 coded-broad-SV BD cases, and 3 952 controls. The estimated h2g were 0.24 (p = 0.015), 0.09 (p = 0.064), 0.13 (p = 0.003), 0.00 (p = 0.591) for 95-NLP, coded-strict, coded-broad and coded-broad-SV BD, respectively. The h2g for all EHR-based cases combined except coded-broad-SV (excluded due to 0 h2g) was 0.12 (p = 0.004). These h2g were lower or similar to the h2g observed by the ICCBD + PGCBD (0.23, p = 3.17E-80, total N = 33,181). However, the rg between ICCBD + PGCBD and the EHR-based cases were high for 95-NLP (0.66, p = 3.69 × 10-5), coded-strict (1.00, p = 2.40 × 10-4), and coded-broad (0.74, p = 8.11 × 10-7). The rg between EHR-based BD definitions ranged from 0.90 to 0.98. These results provide the first genetic validation of automated EHR-based phenotyping for BD and suggest that this approach identifies cases that are highly genetically correlated with those ascertained through conventional methods. High throughput phenotyping using the large data resources available in EHRs represents a viable method for accelerating psychiatric genetic research.
| S-EPMC5904248 | biostudies-literature
Project description:Current studies regarding the secondary use of electronic health records (EHR) predominantly rely on domain expertise and existing medical knowledge. Though significant efforts have been devoted to investigating the application of machine learning algorithms in the EHR, efficient and powerful representation of patients is needed to unleash the potential of discovering new medical patterns underlying the EHR. Here, we present an unsupervised method for embedding high-dimensional EHR data at the patient level, aimed at characterizing patient heterogeneity in complex diseases and identifying new disease patterns associated with clinical outcome disparities. Inspired by the architecture of modern language models-specifically transformers with attention mechanisms, we use patient diagnosis and procedure codes as vocabularies and treat each patient as a sentence to perform the patient embedding. We applied this approach to 34,851 unique medical codes across 1,046,649 longitudinal patient events, including 102,739 patients from the electronic Medical Records and GEnomics (eMERGE) Network. The resulting patient vectors demonstrated excellent performance in predicting future disease events (median AUROC = 0.87 within one year) and bulk phenotyping (median AUROC = 0.84). We then illustrated the utility of these patient vectors in revealing heterogeneous comorbidity patterns, exemplified by disease subtypes in colorectal cancer and systemic lupus erythematosus, and capturing distinct longitudinal disease trajectories. External validation using EHR data from the University of Washington confirmed robust model performance, with median AUROCs of 0.83 and 0.84 for bulk phenotyping tasks and disease onset prediction, respectively. Importantly, the model reproduced the clustering results of disease subtypes identified in the eMERGE cohort and uncovered variations in overall mortality among these subtypes. Together, these results underscore the potential of representation learning in EHRs to enhance patient characterization and associated clinical outcomes, thereby advancing disease forecasting and facilitating personalized medicine.
| S-EPMC11469380 | biostudies-literature
Project description:ObjectiveElectronic Health Record (EHR) based phenotyping is a crucial yet challenging problem in the biomedical field. Though clinicians typically determine patient-level diagnoses via manual chart review, the sheer volume and heterogeneity of EHR data renders such tasks challenging, time-consuming, and prohibitively expensive, thus leading to a scarcity of clinical annotations in EHRs. Weakly supervised learning algorithms have been successfully applied to various EHR phenotyping problems, due to their ability to leverage information from large quantities of unlabeled samples to better inform predictions based on a far smaller number of patients. However, most weakly supervised methods are subject to the challenge to choose the right cutoff value to generate an optimal classifier. Furthermore, since they only utilize the most informative features (i.e., main ICD and NLP counts) they may fail for episodic phenotypes that cannot be consistently detected via ICD and NLP data. In this paper, we propose a label-efficient, weakly semi-supervised deep learning algorithm for EHR phenotyping (WSS-DL), which overcomes the limitations above.Materials and methodsWSS-DL classifies patient-level disease status through a series of learning stages: 1) generating silver standard labels, 2) deriving enhanced-silver-standard labels by fitting a weakly supervised deep learning model to data with silver standard labels as outcomes and high dimensional EHR features as input, and 3) obtaining the final prediction score and classifier by fitting a supervised learning model to data with a minimal number of gold standard labels as the outcome, and the enhanced-silver-standard labels and a minimal set of most informative EHR features as input. To assess the generalizability of WSS-DL across different phenotypes and medical institutions, we apply WSS-DL to classify a total of 17 diseases, including both acute and chronic conditions, using EHR data from three healthcare systems. Additionally, we determine the minimum quantity of training labels required by WSS-DL to outperform existing supervised and semi-supervised phenotyping methods.ResultsThe proposed method, in combining the strengths of deep learning and weakly semi-supervised learning, successfully leverages the crucial phenotyping information contained in EHR features from unlabeled samples. Indeed, the deep learning model's ability to handle high-dimensional EHR features allows it to generate strong phenotype status predictions from silver standard labels. These predictions, in turn, provide highly effective features in the final logistic regression stage, leading to high phenotyping accuracy in notably small subsets of labeled data (e.g. n = 40 labeled samples).ConclusionOur method's high performance in EHR datasets with very small numbers of labels indicates its potential value in aiding doctors to diagnose rare diseases as well as conditions susceptible to misdiagnosis.
| S-EPMC10112494 | biostudies-literature
Project description:Deep learning (DL)-based predictive models from electronic health records (EHRs) deliver impressive performance in many clinical tasks. Large training cohorts, however, are often required by these models to achieve high accuracy, hindering the adoption of DL-based models in scenarios with limited training data. Recently, bidirectional encoder representations from transformers (BERT) and related models have achieved tremendous successes in the natural language processing domain. The pretraining of BERT on a very large training corpus generates contextualized embeddings that can boost the performance of models trained on smaller datasets. Inspired by BERT, we propose Med-BERT, which adapts the BERT framework originally developed for the text domain to the structured EHR domain. Med-BERT is a contextualized embedding model pretrained on a structured EHR dataset of 28,490,650 patients. Fine-tuning experiments showed that Med-BERT substantially improves the prediction accuracy, boosting the area under the receiver operating characteristics curve (AUC) by 1.21-6.14% in two disease prediction tasks from two clinical databases. In particular, pretrained Med-BERT obtains promising performances on tasks with small fine-tuning training sets and can boost the AUC by more than 20% or obtain an AUC as high as a model trained on a training set ten times larger, compared with deep learning models without Med-BERT. We believe that Med-BERT will benefit disease prediction studies with small local training datasets, reduce data collection expenses, and accelerate the pace of artificial intelligence aided healthcare.
| S-EPMC8137882 | biostudies-literature
Project description:In order to advance precision medicine, detailed clinical features ought to be described in a way that leverages current knowledge. Although data collected from biomedical research is expanding at an almost exponential rate, our ability to transform that information into patient care has not kept at pace. A major barrier preventing this transformation is that multi-dimensional data collection and analysis is usually carried out without much understanding of the underlying knowledge structure. Here, in an effort to bridge this gap, Electronic Health Records (EHRs) of individual patients are connected to a heterogeneous knowledge network called Scalable Precision Medicine Oriented Knowledge Engine (SPOKE). Then an unsupervised machine-learning algorithm creates Propagated SPOKE Entry Vectors (PSEVs) that encode the importance of each SPOKE node for any code in the EHRs. We argue that these results, alongside the natural integration of PSEVs into any EHR machine-learning platform, provide a key step toward precision medicine.
| S-EPMC6620318 | biostudies-literature
Project description:Longitudinal electronic health records (EHR) can be utilized to identify patterns of disease development and progression in real-world settings. Unsupervised temporal matching algorithms are being repurposed to EHR from signal processing- and protein-sequence alignment tasks where they have shown immense promise for gaining insight into disease. The robustness of these algorithms for classifying EHR clinical data remains to be determined. Timeseries compiled from clinical measurements, such as blood pressure, have far more irregularity in sampling and missingness than the data for which these algorithms were developed, necessitating a systematic evaluation of these methods. We applied 30 state-of-the-art unsupervised machine learning algorithms to 6,912 systematically generated simulated clinical datasets across five parameters. These algorithms included eight temporal matching algorithms with fourteen partitional and eight fuzzy clustering methods. Nemenyi tests were used to determine differences in accuracy using the Adjusted Rand Index (ARI). Dynamic time warping and its lower-bound variants had the highest accuracies across all cohorts (median ARI>0.70). All 30 methods were better at discriminating classes with differences in magnitude compared to differences in trajectory shapes. Missingness impacted accuracies only when classes were different by trajectory shape. The method with the highest ARI was then used to cluster a large pediatric metabolic syndrome (MetS) cohort (N = 43,426). We identified three unique childhood BMI patterns with high average cluster consensus (>70%). The algorithm identified a cluster with consistently high BMI which had the greatest risk of MetS, consistent with prior literature (OR = 4.87, 95% CI: 3.93-6.12). While these algorithms have been shown to have similar accuracies for regular timeseries, their accuracies in clinical applications vary substantially in discriminating differences in shape and especially with moderate to high missingness (>10%). This systematic assessment also shows that the most robust algorithms tested here can derive meaningful insights from longitudinal clinical data.
| S-EPMC11478862 | biostudies-literature
Project description:BackgroundIdentifying pregnancies at risk for preterm birth, one of the leading causes of worldwide infant mortality, has the potential to improve prenatal care. However, we lack broadly applicable methods to accurately predict preterm birth risk. The dense longitudinal information present in electronic health records (EHRs) is enabling scalable and cost-efficient risk modeling of many diseases, but EHR resources have been largely untapped in the study of pregnancy.MethodsHere, we apply machine learning to diverse data from EHRs with 35,282 deliveries to predict singleton preterm birth.ResultsWe find that machine learning models based on billing codes alone can predict preterm birth risk at various gestational ages (e.g., ROC-AUC = 0.75, PR-AUC = 0.40 at 28 weeks of gestation) and outperform comparable models trained using known risk factors (e.g., ROC-AUC = 0.65, PR-AUC = 0.25 at 28 weeks). Examining the patterns learned by the model reveals it stratifies deliveries into interpretable groups, including high-risk preterm birth subtypes enriched for distinct comorbidities. Our machine learning approach also predicts preterm birth subtypes (spontaneous vs. indicated), mode of delivery, and recurrent preterm birth. Finally, we demonstrate the portability of our approach by showing that the prediction models maintain their accuracy on a large, independent cohort (5978 deliveries) from a different healthcare system.ConclusionsBy leveraging rich phenotypic and genetic features derived from EHRs, we suggest that machine learning algorithms have great potential to improve medical care during pregnancy. However, further work is needed before these models can be applied in clinical settings.
| S-EPMC9516830 | biostudies-literature
Project description:Electronic health records (EHR) provide opportunities for deeper understanding of human phenotypes - in our case, latent disease - based on statistical modeling. We propose a categorical matrix factorization method to infer latent diseases from EHR data. A latent disease is defined as an unknown biological aberration that causes a set of common symptoms for a group of patients. The proposed approach is based on a novel double feature allocation model which simultaneously allocates features to the rows and the columns of a categorical matrix. Using a Bayesian approach, available prior information on known diseases (e.g., hypertension and diabetes) greatly improves identifiability and interpretability of the latent diseases. We assess the proposed approach by simulation studies including mis-specified models and comparison with sparse latent factor models. In the application to a Chinese EHR data set, we identify 10 latent diseases, each of which is shared by groups of subjects with specific health traits related to lipid disorder, thrombocytopenia, polycythemia, anemia, bacterial and viral infections, allergy, and malnutrition. The identification of the latent diseases can help healthcare officials better monitor the subjects' ongoing health conditions and look into potential risk factors and approaches for disease prevention. We cross-check the reported latent diseases with medical literature and find agreement between our discovery and reported findings elsewhere. We provide an R package "dfa" implementing our method and an R shiny web application reporting the findings.
| S-EPMC10727496 | biostudies-literature
Project description:Electronic health record (EHR) data are increasingly used to support real-world evidence studies but are limited by the lack of precise timings of clinical events. Here, we propose a label-efficient incident phenotyping (LATTE) algorithm to accurately annotate the timing of clinical events from longitudinal EHR data. By leveraging the pre-trained semantic embeddings, LATTE selects predictive features and compresses their information into longitudinal visit embeddings through visit attention learning. LATTE models the sequential dependency between the target event and visit embeddings to derive the timings. To improve label efficiency, LATTE constructs longitudinal silver-standard labels from unlabeled patients to perform semi-supervised training. LATTE is evaluated on the onset of type 2 diabetes, heart failure, and relapses of multiple sclerosis. LATTE consistently achieves substantial improvements over benchmark methods while providing high prediction interpretability. The event timings are shown to help discover risk factors of heart failure among patients with rheumatoid arthritis.
| S-EPMC10801250 | biostudies-literature