Toward high-throughput phenotyping: unbiased automated feature extraction and selection from knowledge sources

Associations of APOE variants with sphingomyelin and cholesterol metabolites across the life-course in diverse populations

INTRODUCTION

Two alleles (ε2 and ε4) in the APOE gene are known to be strongly associated with lipid metabolism disorders, such as dyslipidemia, which are important risk factors for the development of cardiovascular diseases. While prior research has largely centered on adult populations, establishing APOE-lipid associations in infants, children, and adolescents-especially those from historically understudied groups-could inform earlier interventions and treatments.

OBJECTIVES

This study aimed to evaluate the dependence of the metabolome on the APOE variants using five diverse cohorts that span infancy through adulthood, comprising a total of over 190,000 individuals.

METHODS

We extracted the APOE variants (rs7412 and rs429358) from all cohorts-testing both the ε2 allele (rs7412-T and rs429358-T) and the ε4 allele (rs7412-C and rs429358-C)-and evaluated their associations with the global plasma metabolome which was measured by mass spectrometry-based (Metabolon or Broad Institute) or NMR-based (Nightingale) assays depending on the cohort, using a Bonferroni-corrected significance threshold.

RESULTS

Among 589 metabolites tested in our discovery population, only six including sphingomyelins and cholesterol were significantly associated with the rs7412/ε2 allele. Sphingomyelin (d18:1/22:0) and cholesterol were negatively associated with ε2 allele (β-value = -0.54 [-0.76, -0.32] p-value = 1.39 × 10-6 and - 0.55 [-0.77, -0.33]; p-value = 1.49 × 10-6, respectively). These relationships were replicated in the four additional cohorts without heterogeneity.

CONCLUSION

Our findings support the need for early intervention in lipid levels regardless of age, sex, and ethnicity and further investigations of the APOE variants on risk of various diseases in later life.

CATI: A medical context-enhanced framework for diagnosis code assignment in the UK Biobank study

Diagnosis codes are standard code format of diseases or medical conditions. This study is aimed at assigning diagnosis codes to patients in large-scale biobanks, particularly addressing the issue of missing codes for some patients. This is crucial for downstream disease-related tasks. While recent methods primarily rely on structured biobank data for code assignment, they often overlook the valuable medical context provided by textual information in the biobanks and hierarchical structure of the disease coding system. To address this gap, we have developed CATI, a medical context-enhanced framework for diagnosis Code Assignment by integrating Textual details derived from key features and disease hIerarchy. The study is based on the UK Biobank data and considers Phecodes and ICD-10 codes as standard disease formats. We start by representing ten informative codified features using their formal names and then integrate them into CATI as text embeddings, achieved through prompt tuning on the pre-trained language model BioBERT. Recognizing the hierarchical structure of diagnosis codes, we have developed a novel convolution layer in our method that effectively propagates logits between adjacent diagnosis codes. Evaluation results demonstrate that CATI outperforms existing state-of-the-art methods in terms of both Phecodes and ICD-10 codes, boasting at least a 5.16% improvement in average AUROC for unseen disease codes and an 8.68% rise in average AUPRC for disease codes with training instances ranging in (1000,10000]. This framework contributes to the formation of well-defined cohorts for downstream studies and offers a unique perspective for addressing complex healthcare tasks by incorporating vital medical context.

Rare variant analyses in 51,256 type 2 diabetes cases and 370,487 controls reveal the pathogenicity spectrum of monogenic diabetes genes

Type 2 diabetes (T2D) genome-wide association studies (GWASs) often overlook rare variants as a result of previous imputation panels' limitations and scarce whole-genome sequencing (WGS) data. We used TOPMed imputation and WGS to conduct the largest T2D GWAS meta-analysis involving 51,256 cases of T2D and 370,487 controls, targeting variants with a minor allele frequency as low as 5 × 10-5. We identified 12 new variants, including a rare African/African American-enriched enhancer variant near the LEP gene (rs147287548), associated with fourfold increased T2D risk. We also identified a rare missense variant in HNF4A (p.Arg114Trp), associated with eightfold increased T2D risk, previously reported in maturity-onset diabetes of the young with reduced penetrance, but observed here in a T2D GWAS. We further leveraged these data to analyze 1,634 ClinVar variants in 22 genes related to monogenic diabetes, identifying two additional rare variants in HNF1A and GCK associated with fivefold and eightfold increased T2D risk, respectively, the effects of which were modified by the individual's polygenic risk score. For 21% of the variants with conflicting interpretations or uncertain significance in ClinVar, we provided support of being benign based on their lack of association with T2D. Our work provides a framework for using rare variant GWASs to identify large-effect variants and assess variant pathogenicity in monogenic diabetes genes.

The association of TNF inhibitor use with incident cardiovascular events in radiographic axial spondyloarthritis

BACKGROUND

Whether tumor necrosis factor inhibitor (TNFi) use is cardioprotective among individuals with radiographic axial spondyloarthritis (r-axSpA), who have heightened cardiovascular (CV) risk, is unclear. We tested the association of TNFi use with incident CV outcomes in r-axSpA.

METHODS

We identified a r-axSpA cohort within a Veterans Affairs database between 2002 and 2019 using novel phenotyping methods and secondarily using ICD codes. TNFi use was assessed as a time-varying exposure using pharmacy dispense records. The primary outcome was incident CV disease identified using ICD codes for coronary artery disease, myocardial infarction or stroke. We fit Cox models with inverse probability weights to estimate the risk of each outcome with TNFi use versus non-use. Analyses were performed in the overall cohort, and separately in two periods (2002-2010, 2011-2019) to account for secular trends.

RESULTS

Using phenotyping we identified 26,928 individuals with an r-axSpA diagnosis (mean age 63.4 years, 94 % male); at baseline 3633 were TNFi users and 23,295 were non-users. During follow-up of a mean 3.3 ± 4.2 years, 674 (18.6 %) TNFi users had incident CVD versus 11,838 (50.8 %) non-users. In adjusted analyses, TNFi use versus non-use was associated with lower risk of incident CVD (HR 0.34, 95 % CI 0.29-0.40) in the cohort overall, and in the two time periods separately.

CONCLUSION

In this r-axSpA cohort identified using phenotyping methods, TNFi use versus non-use had a lower risk of incident CVD. These findings provide reassurance regarding the CV safety of TNFi agents for r-axSpA treatment. Replication of these results in other cohorts is needed.

Protocol for Designing a Model to Predict the Likelihood of Psychosis From Electronic Health Records Using Natural Language Processing and Machine Learning

INTRODUCTION

Rapid identification of individuals developing a psychotic spectrum disorder (PSD) is crucial because untreated psychosis is associated with poor outcomes and decreased treatment response. Lack of recognition of early psychotic symptoms often delays diagnosis, further worsening these outcomes.

METHODS

The proposed study is a cross-sectional, retrospective analysis of electronic health record data including clinician documentation and patient-clinician secure messages for patients aged 15-29 years with ≥ 1 primary care encounter between 2017 and 2019 within 2 Kaiser Permanente regions. Patients with new-onset PSD will be distinguished from those without a diagnosis if they have ≥ 1 PSD diagnosis within 12 months following the primary care encounter. The prediction model will be trained using a trisourced natural language processing feature extraction design and validated both within each region separately and in a modified combined sample.

DISCUSSION

This proposed model leverages the strengths of the large volume of patient-specific data from an integrated electronic health record with natural language processing to identify patients at elevated chance of developing a PSD. This project carries the potential to reduce the duration of untreated psychosis and thereby improve long-term patient outcomes.

Semi-supervised Double Deep Learning Temporal Risk Prediction (SeDDLeR) with Electronic Health Records

BACKGROUND

Risk prediction plays a crucial role in planning for prevention, monitoring, and treatment. Electronic Health Records (EHRs) offer an expansive repository of temporal medical data encompassing both risk factors and outcome indicators essential for effective risk prediction. However, challenges emerge due to the lack of readily available gold-standard outcomes and the complex effects of various risk factors. Compounding these challenges are the false positives in diagnosis codes, and formidable task of pinpointing the onset timing in annotations.

OBJECTIVE

We develop a Semi-supervised Double Deep Learning Temporal Risk Prediction (SeDDLeR) algorithm based on extensive unlabeled longitudinal Electronic Health Records (EHR) data augmented by a limited set of gold standard labels on the binary status information indicating whether the clinical event of interest occurred during the follow-up period.

METHODS

The SeDDLeR algorithm calculates an individualized risk of developing future clinical events over time using each patient's baseline EHR features via the following steps: (1) construction of an initial EHR-derived surrogate as a proxy for the onset status; (2) deep learning calibration of the surrogate along gold-standard onset status; and (3) semi-supervised deep learning for risk prediction combining calibrated surrogates and gold-standard onset status. To account for missing onset time and heterogeneous follow-up, we introduce temporal kernel weighting. We devise a Gated Recurrent Units (GRUs) module to capture temporal characteristics. We subsequently assess our proposed SeDDLeR method in simulation studies and apply the method to the Massachusetts General Brigham (MGB) Biobank to predict type 2 diabetes (T2D) risk.

RESULTS

SeDDLeR outperforms benchmark risk prediction methods, including Semi-parametric Transformation Model (STM) and DeepHit, with consistently best accuracy across experiments. SeDDLeR achieved the best C-statistics ( 0.815, SE 0.023; vs STM +.084, SE 0.030, P-value .004; vs DeepHit +.055, SE 0.027, P-value .024) and best average time-specific AUC (0.778, SE 0.022; vs STM + 0.059, SE 0.039, P-value .067; vs DeepHit + 0.168, SE 0.032, P-value <0.001) in the MGB T2D study.

CONCLUSION

SeDDLeR can train robust risk prediction models in both real-world EHR and synthetic datasets with minimal requirements of labeling event times. It holds the potential to be incorporated for future clinical trial recruitment or clinical decision-making.

Machine learning-derived phenotypic trajectories of asthma and allergy in children and adolescents: protocol for a systematic review

INTRODUCTION

Development of asthma and allergies in childhood/adolescence commonly follows a sequential progression termed the 'atopic march'. Recent reports indicate, however, that these diseases are composed of multiple distinct phenotypes, with possibly differential trajectories. We aim to synthesise the current literature in the field of machine learning-based trajectory studies of asthma/allergies in children and adolescents, summarising the frequency, characteristics and associated risk factors and outcomes of identified trajectories and indicating potential directions for subsequent research in replicability, pathophysiology, risk stratification and personalised management. Furthermore, methodological approaches and quality will be critically appraised, highlighting trends, limitations and future perspectives.

METHODS AND ANALYSES

10 databases (CAB Direct, CINAHL, Embase, Google Scholar, PsycInfo, PubMed, Scopus, Web of Science, WHO Global Index Medicus and WorldCat Dissertations and Theses) will be searched for observational studies (including conference abstracts and grey literature) from the last 10 years (2013-2023) without restriction by language. Screening, data extraction and assessment of quality and risk of bias (using a custom-developed tool) will be performed independently in pairs. The characteristics of the derived trajectories will be narratively synthesised, tabulated and visualised in figures. Risk factors and outcomes associated with the trajectories will be summarised and pooled estimates from comparable numerical data produced through random-effects meta-analysis. Methodological approaches will be narratively synthesised and presented in tabulated form and figure to visualise trends.

ETHICS AND DISSEMINATION

Ethical approval is not warranted as no patient-level data will be used. The findings will be published in an international peer-reviewed journal.

PROSPERO REGISTRATION NUMBER

CRD42023441691.

A general framework for developing computable clinical phenotype algorithms

OBJECTIVE

To present a general framework providing high-level guidance to developers of computable algorithms for identifying patients with specific clinical conditions (phenotypes) through a variety of approaches, including but not limited to machine learning and natural language processing methods to incorporate rich electronic health record data.

MATERIALS AND METHODS

Drawing on extensive prior phenotyping experiences and insights derived from 3 algorithm development projects conducted specifically for this purpose, our team with expertise in clinical medicine, statistics, informatics, pharmacoepidemiology, and healthcare data science methods conceptualized stages of development and corresponding sets of principles, strategies, and practical guidelines for improving the algorithm development process.

RESULTS

We propose 5 stages of algorithm development and corresponding principles, strategies, and guidelines: (1) assessing fitness-for-purpose, (2) creating gold standard data, (3) feature engineering, (4) model development, and (5) model evaluation.

DISCUSSION AND CONCLUSION

This framework is intended to provide practical guidance and serve as a basis for future elaboration and extension.

Semi-supervised ROC analysis for reliable and streamlined evaluation of phenotyping algorithms

OBJECTIVE

High-throughput phenotyping will accelerate the use of electronic health records (EHRs) for translational research. A critical roadblock is the extensive medical supervision required for phenotyping algorithm (PA) estimation and evaluation. To address this challenge, numerous weakly-supervised learning methods have been proposed. However, there is a paucity of methods for reliably evaluating the predictive performance of PAs when a very small proportion of the data is labeled. To fill this gap, we introduce a semi-supervised approach (ssROC) for estimation of the receiver operating characteristic (ROC) parameters of PAs (eg, sensitivity, specificity).

MATERIALS AND METHODS

ssROC uses a small labeled dataset to nonparametrically impute missing labels. The imputations are then used for ROC parameter estimation to yield more precise estimates of PA performance relative to classical supervised ROC analysis (supROC) using only labeled data. We evaluated ssROC with synthetic, semi-synthetic, and EHR data from Mass General Brigham (MGB).

RESULTS

ssROC produced ROC parameter estimates with minimal bias and significantly lower variance than supROC in the simulated and semi-synthetic data. For the 5 PAs from MGB, the estimates from ssROC are 30% to 60% less variable than supROC on average.

DISCUSSION

ssROC enables precise evaluation of PA performance without demanding large volumes of labeled data. ssROC is also easily implementable in open-source R software.

CONCLUSION

When used in conjunction with weakly-supervised PAs, ssROC facilitates the reliable and streamlined phenotyping necessary for EHR-based research.

Data-driven automated classification algorithms for acute health conditions: applying PheNorm to COVID-19 disease

OBJECTIVES

Automated phenotyping algorithms can reduce development time and operator dependence compared to manually developed algorithms. One such approach, PheNorm, has performed well for identifying chronic health conditions, but its performance for acute conditions is largely unknown. Herein, we implement and evaluate PheNorm applied to symptomatic COVID-19 disease to investigate its potential feasibility for rapid phenotyping of acute health conditions.

MATERIALS AND METHODS

PheNorm is a general-purpose automated approach to creating computable phenotype algorithms based on natural language processing, machine learning, and (low cost) silver-standard training labels. We applied PheNorm to cohorts of potential COVID-19 patients from 2 institutions and used gold-standard manual chart review data to investigate the impact on performance of alternative feature engineering options and implementing externally trained models without local retraining.

RESULTS

Models at each institution achieved AUC, sensitivity, and positive predictive value of 0.853, 0.879, 0.851 and 0.804, 0.976, and 0.885, respectively, at quantiles of model-predicted risk that maximize F1. We report performance metrics for all combinations of silver labels, feature engineering options, and models trained internally versus externally.

DISCUSSION

Phenotyping algorithms developed using PheNorm performed well at both institutions. Performance varied with different silver-standard labels and feature engineering options. Models developed locally at one site also worked well when implemented externally at the other site.

CONCLUSION

PheNorm models successfully identified an acute health condition, symptomatic COVID-19. The simplicity of the PheNorm approach allows it to be applied at multiple study sites with substantially reduced overhead compared to traditional approaches.

Machine Learning Informed Diagnosis for Congenital Heart Disease in Large Claims Data Source

Background

With an increasing interest in using large claims databases in medical practice and research, it is a meaningful and essential step to efficiently identify patients with the disease of interest.

Objectives

This study aims to establish a machine learning (ML) approach to identify patients with congenital heart disease (CHD) in large claims databases.

Methods

We harnessed data from the Quebec claims and hospitalization databases from 1983 to 2000. The study included 19,187 patients. Of them, 3,784 were labeled as true CHD patients using a clinician developed algorithm with manual audits considered as the gold standards. To establish an accurate ML-empowered automated CHD classification system, we evaluated ML methods including Gradient Boosting Decision Tree, Support Vector Machine, Decision tree, and compared them to regularized logistic regression. The Area Under the Precision Recall Curve was used as the evaluation metric. External validation was conducted with an updated data set to 2010 with different subjects.

Results

Among the ML methods we evaluated, Gradient Boosting Decision Tree led the performance in identifying true CHD patients with 99.3% Area Under the Precision Recall Curve, 98.0% for sensitivity, and 99.7% for specificity. External validation returned similar statistics on model performance.

Conclusions

This study shows that a tedious and time-consuming clinical inspection for CHD patient identification can be replaced by an extremely efficient ML algorithm in large claims database. Our findings demonstrate that ML methods can be used to automate complicated algorithms to identify patients with complex diseases.

Clinical concept annotation with contextual word embedding in active transfer learning environment

Objective

The study aims to present an active learning approach that automatically extracts clinical concepts from unstructured data and classifies them into explicit categories such as Problem, Treatment, and Test while preserving high precision and recall and demonstrating the approach through experiments using i2b2 public datasets.

Methods

Initially labeled data are acquired from a lexical-based approach in sufficient amounts to perform an active learning process. A contextual word embedding similarity approach is adopted using BERT base variant models such as ClinicalBERT, DistilBERT, and SCIBERT to automatically classify the unlabeled clinical concept into explicit categories. Additionally, deep learning and large language model (LLM) are trained on acquiring label data through active learning.

Results

Using i2b2 datasets (426 clinical notes), the lexical-based method achieved precision, recall, and F1-scores of 76%, 70%, and 73%. SCIBERT excelled in active transfer learning, yielding precision of 70.84%, recall of 77.40%, F1-score of 73.97%, and accuracy of 69.30%, surpassing counterpart models. Among deep learning models, convolutional neural networks (CNNs) trained with embeddings (BERTBase, DistilBERT, SCIBERT, ClinicalBERT) achieved training accuracies of 92-95% and testing accuracies of 89-93%. These results were higher compared to other deep learning models. Additionally, we individually evaluated these LLMs; among them, ClinicalBERT achieved the highest performance, with a training accuracy of 98.4% and a testing accuracy of 96%, outperforming the others.

Conclusions

The proposed methodology enhances clinical concept extraction by integrating active learning and models like SCIBERT and CNN. It improves annotation efficiency while maintaining high accuracy, showcasing potential for clinical applications.

Zero-shot interpretable phenotyping of postpartum hemorrhage using large language models

Many areas of medicine would benefit from deeper, more accurate phenotyping, but there are limited approaches for phenotyping using clinical notes without substantial annotated data. Large language models (LLMs) have demonstrated immense potential to adapt to novel tasks with no additional training by specifying task-specific instructions. Here we report the performance of a publicly available LLM, Flan-T5, in phenotyping patients with postpartum hemorrhage (PPH) using discharge notes from electronic health records (n = 271,081). The language model achieves strong performance in extracting 24 granular concepts associated with PPH. Identifying these granular concepts accurately allows the development of interpretable, complex phenotypes and subtypes. The Flan-T5 model achieves high fidelity in phenotyping PPH (positive predictive value of 0.95), identifying 47% more patients with this complication compared to the current standard of using claims codes. This LLM pipeline can be used reliably for subtyping PPH and outperforms a claims-based approach on the three most common PPH subtypes associated with uterine atony, abnormal placentation, and obstetric trauma. The advantage of this approach to subtyping is its interpretability, as each concept contributing to the subtype determination can be evaluated. Moreover, as definitions may change over time due to new guidelines, using granular concepts to create complex phenotypes enables prompt and efficient updating of the algorithm. Using this language modelling approach enables rapid phenotyping without the need for any manually annotated training data across multiple clinical use cases.

A Type 1 Diabetes Polygenic Score Is Not Associated With Prevalent Type 2 Diabetes in Large Population Studies

Context

Both type 1 diabetes (T1D) and type 2 diabetes (T2D) have significant genetic contributions to risk and understanding their overlap can offer clinical insight.

Objective

We examined whether a T1D polygenic score (PS) was associated with a diagnosis of T2D in the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium.

Methods

We constructed a T1D PS using 79 known single nucleotide polymorphisms associated with T1D risk. We analyzed 13 792 T2D cases and 14 169 controls from CHARGE cohorts to determine the association between the T1D PS and T2D prevalence. We validated findings in an independent sample of 2256 T2D cases and 27 052 controls from the Mass General Brigham Biobank (MGB Biobank). As secondary analyses in 5228 T2D cases from CHARGE, we used multivariable regression models to assess the association of the T1D PS with clinical outcomes associated with T1D.

Results

The T1D PS was not associated with T2D both in CHARGE (P = .15) and in the MGB Biobank (P = .87). The partitioned human leukocyte antigens only PS was associated with T2D in CHARGE (OR 1.02 per 1 SD increase in PS, 95% CI 1.01-1.03, P = .006) but not in the MGB Biobank. The T1D PS was weakly associated with insulin use (OR 1.007, 95% CI 1.001-1.012, P = .03) in CHARGE T2D cases but not with other outcomes.

Conclusion

In large biobank samples, a common variant PS for T1D was not consistently associated with prevalent T2D. However, possible heterogeneity in T2D cannot be ruled out and future studies are needed do subphenotyping.

Rare variant association analysis in 51,256 type 2 diabetes cases and 370,487 controls informs the spectrum of pathogenicity of monogenic diabetes genes

We meta-analyzed array data imputed with the TOPMed reference panel and whole-genome sequence (WGS) datasets and performed the largest, rare variant (minor allele frequency as low as 5×10-5) GWAS meta-analysis of type 2 diabetes (T2D) comprising 51,256 cases and 370,487 controls. We identified 52 novel variants at genome-wide significance (p<5 × 10-8), including 8 novel variants that were either rare or ancestry-specific. Among them, we identified a rare missense variant in HNF4A p.Arg114Trp (OR=8.2, 95% confidence interval [CI]=4.6-14.0, p = 1.08×10-13), previously reported as a variant implicated in Maturity Onset Diabetes of the Young (MODY) with incomplete penetrance. We demonstrated that the diabetes risk in carriers of this variant was modulated by a T2D common variant polygenic risk score (cvPRS) (carriers in the top PRS tertile [OR=18.3, 95%CI=7.2-46.9, p=1.2×10-9] vs carriers in the bottom PRS tertile [OR=2.6, 95% CI=0.97-7.09, p = 0.06]. Association results identified eight variants of intermediate penetrance (OR>5) in monogenic diabetes (MD), which in aggregate as a rare variant PRS were associated with T2D in an independent WGS dataset (OR=4.7, 95% CI=1.86-11.77], p = 0.001). Our data also provided support evidence for 21% of the variants reported in ClinVar in these MD genes as benign based on lack of association with T2D. Our work provides a framework for using rare variant imputation and WGS analyses in large-scale population-based association studies to identify large-effect rare variants and provide evidence for informing variant pathogenicity.

Discovering cellular programs of intrinsic and extrinsic drivers of metabolic traits using LipocyteProfiler

A primary obstacle in translating genetic associations with disease into therapeutic strategies is elucidating the cellular programs affected by genetic risk variants and effector genes. Here, we introduce LipocyteProfiler, a cardiometabolic-disease-oriented high-content image-based profiling tool that enables evaluation of thousands of morphological and cellular profiles that can be systematically linked to genes and genetic variants relevant to cardiometabolic disease. We show that LipocyteProfiler allows surveillance of diverse cellular programs by generating rich context- and process-specific cellular profiles across hepatocyte and adipocyte cell-state transitions. We use LipocyteProfiler to identify known and novel cellular mechanisms altered by polygenic risk of metabolic disease, including insulin resistance, fat distribution, and the polygenic contribution to lipodystrophy. LipocyteProfiler paves the way for large-scale forward and reverse deep phenotypic profiling in lipocytes and provides a framework for the unbiased identification of causal relationships between genetic variants and cellular programs relevant to human disease.

Zero-shot Interpretable Phenotyping of Postpartum Hemorrhage Using Large Language Models

Many areas of medicine would benefit from deeper, more accurate phenotyping, but there are limited approaches for phenotyping using clinical notes without substantial annotated data. Large language models (LLMs) have demonstrated immense potential to adapt to novel tasks with no additional training by specifying task-specific i nstructions. We investigated the per-formance of a publicly available LLM, Flan-T5, in phenotyping patients with postpartum hemorrhage (PPH) using discharge notes from electronic health records ( n =271,081). The language model achieved strong performance in extracting 24 granular concepts associated with PPH. Identifying these granular concepts accurately allowed the development of inter-pretable, complex phenotypes and subtypes. The Flan-T5 model achieved high fidelity in phenotyping PPH (positive predictive value of 0.95), identifying 47% more patients with this complication compared to the current standard of using claims codes. This LLM pipeline can be used reliably for subtyping PPH and outperformed a claims-based approach on the three most common PPH subtypes associated with uterine atony, abnormal placentation, and obstetric trauma. The advantage of this approach to subtyping is its interpretability, as each concept contributing to the subtype determination can be evaluated. Moreover, as definitions may change over time due to new guidelines, using granular concepts to create complex phenotypes enables prompt and efficient updating of the algorithm. Using this lan-guage modelling approach enables rapid phenotyping without the need for any manually annotated training data across multiple clinical use cases.

Temporal characterization of Alzheimer's Disease with sequences of clinical records

BACKGROUND

Alzheimer's Disease (AD) is a complex clinical phenotype with unprecedented social and economic tolls on an ageing global population. Real-world data (RWD) from electronic health records (EHRs) offer opportunities to accelerate precision drug development and scale epidemiological research on AD. A precise characterization of AD cohorts is needed to address the noise abundant in RWD.

METHODS

We conducted a retrospective cohort study to develop and test computational models for AD cohort identification using clinical data from 8 Massachusetts healthcare systems. We mined temporal representations from EHR data using the transitive sequential pattern mining algorithm (tSPM) to train and validate our models. We then tested our models against a held-out test set from a review of medical records to adjudicate the presence of AD. We trained two classes of Machine Learning models, using Gradient Boosting Machine (GBM), to compare the utility of AD diagnosis records versus the tSPM temporal representations (comprising sequences of diagnosis and medication observations) from electronic medical records for characterizing AD cohorts.

FINDINGS

In a group of 4985 patients, we identified 219 tSPM temporal representations (i.e., transitive sequences) of medical records for constructing the best classification models. The models with sequential features improved AD classification by a magnitude of 3-16 percent over the use of AD diagnosis codes alone. The computed cohort included 663 patients, 35 of whom had no record of AD. Six groups of tSPM sequences were identified for characterizing the AD cohorts.

INTERPRETATION

We present sequential patterns of diagnosis and medication codes from electronic medical records, as digital markers of Alzheimer's Disease. Classification algorithms developed on sequential patterns can replace standard features from EHRs to enrich phenotype modelling.

FUNDING

National Institutes of Health: the National Institute on Aging (RF1AG074372) and the National Institute of Allergy and Infectious Diseases (R01AI165535).

Generate Analysis-Ready Data for Real-world Evidence: Tutorial for Harnessing Electronic Health Records With Advanced Informatic Technologies

Although randomized controlled trials (RCTs) are the gold standard for establishing the efficacy and safety of a medical treatment, real-world evidence (RWE) generated from real-world data has been vital in postapproval monitoring and is being promoted for the regulatory process of experimental therapies. An emerging source of real-world data is electronic health records (EHRs), which contain detailed information on patient care in both structured (eg, diagnosis codes) and unstructured (eg, clinical notes and images) forms. Despite the granularity of the data available in EHRs, the critical variables required to reliably assess the relationship between a treatment and clinical outcome are challenging to extract. To address this fundamental challenge and accelerate the reliable use of EHRs for RWE, we introduce an integrated data curation and modeling pipeline consisting of 4 modules that leverage recent advances in natural language processing, computational phenotyping, and causal modeling techniques with noisy data. Module 1 consists of techniques for data harmonization. We use natural language processing to recognize clinical variables from RCT design documents and map the extracted variables to EHR features with description matching and knowledge networks. Module 2 then develops techniques for cohort construction using advanced phenotyping algorithms to both identify patients with diseases of interest and define the treatment arms. Module 3 introduces methods for variable curation, including a list of existing tools to extract baseline variables from different sources (eg, codified, free text, and medical imaging) and end points of various types (eg, death, binary, temporal, and numerical). Finally, module 4 presents validation and robust modeling methods, and we propose a strategy to create gold-standard labels for EHR variables of interest to validate data curation quality and perform subsequent causal modeling for RWE. In addition to the workflow proposed in our pipeline, we also develop a reporting guideline for RWE that covers the necessary information to facilitate transparent reporting and reproducibility of results. Moreover, our pipeline is highly data driven, enhancing study data with a rich variety of publicly available information and knowledge sources. We also showcase our pipeline and provide guidance on the deployment of relevant tools by revisiting the emulation of the Clinical Outcomes of Surgical Therapy Study Group Trial on laparoscopy-assisted colectomy versus open colectomy in patients with early-stage colon cancer. We also draw on existing literature on EHR emulation of RCTs together with our own studies with the Mass General Brigham EHR.

Novel genetic variants associated with inhaled corticosteroid treatment response in older adults with asthma

INTRODUCTION

Older adults have the greatest burden of asthma and poorest outcomes. The pharmacogenetics of inhaled corticosteroid (ICS) treatment response is not well studied in older adults.

METHODS

A genome-wide association study of ICS response was performed in asthmatics of European ancestry in Genetic Epidemiology Research on Adult Health and Aging (GERA) by fitting Cox proportional hazards regression models, followed by validation in the Mass General Brigham (MGB) Biobank and Rotterdam Study. ICS response was measured using two definitions in asthmatics on ICS treatment: (1) absence of oral corticosteroid (OCS) bursts using prescription records and (2) absence of asthma-related exacerbations using diagnosis codes. A fixed-effect meta-analysis was performed for each outcome. The validated single-nucleotide polymorphisms (SNPs) were functionally annotated to standard databases.

RESULTS

In 5710 subjects in GERA, 676 subjects in MGB Biobank, and 465 subjects in the Rotterdam Study, four novel SNPs on chromosome six near PTCHD4 validated across all cohorts and met genome-wide significance on meta-analysis for the OCS burst outcome. In 4541 subjects in GERA and 505 subjects in MGB Biobank, 152 SNPs with p<5 × 10-5 were validated across these two cohorts for the asthma-related exacerbation outcome. The validated SNPs included methylation and expression quantitative trait loci for CPED1, CRADD and DST for the OCS burst outcome and GM2A, SNW1, CACNA1C, DPH1, and RPS10 for the asthma-related exacerbation outcome.

CONCLUSIONS

Multiple novel SNPs associated with ICS response were identified in older adult asthmatics. Several SNPs annotated to genes previously associated with asthma and other airway or allergic diseases, including PTCHD4.

Trends and opportunities in computable clinical phenotyping: A scoping review

Identifying patient cohorts meeting the criteria of specific phenotypes is essential in biomedicine and particularly timely in precision medicine. Many research groups deliver pipelines that automatically retrieve and analyze data elements from one or more sources to automate this task and deliver high-performing computable phenotypes. We applied a systematic approach based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines to conduct a thorough scoping review on computable clinical phenotyping. Five databases were searched using a query that combined the concepts of automation, clinical context, and phenotyping. Subsequently, four reviewers screened 7960 records (after removing over 4000 duplicates) and selected 139 that satisfied the inclusion criteria. This dataset was analyzed to extract information on target use cases, data-related topics, phenotyping methodologies, evaluation strategies, and portability of developed solutions. Most studies supported patient cohort selection without discussing the application to specific use cases, such as precision medicine. Electronic Health Records were the primary source in 87.1 % (N = 121) of all studies, and International Classification of Diseases codes were heavily used in 55.4 % (N = 77) of all studies, however, only 25.9 % (N = 36) of the records described compliance with a common data model. In terms of the presented methods, traditional Machine Learning (ML) was the dominant method, often combined with natural language processing and other approaches, while external validation and portability of computable phenotypes were pursued in many cases. These findings revealed that defining target use cases precisely, moving away from sole ML strategies, and evaluating the proposed solutions in the real setting are essential opportunities for future work. There is also momentum and an emerging need for computable phenotyping to support clinical and epidemiological research and precision medicine.

Drug-related causes attributed to acute kidney injury and their documentation in intensive care patients

PURPOSE

To investigate drug-related causes attributed to acute kidney injury (DAKI) and their documentation in patients admitted to the Intensive Care Unit (ICU).

METHODS

This study was conducted in an academic hospital in the Netherlands by reusing electronic health record (EHR) data of adult ICU admissions between November 2015 to January 2020. First, ICU admissions with acute kidney injury (AKI) stage 2 or 3 were identified. Subsequently, three modes of DAKI documentation in EHR were examined: diagnosis codes (structured data), allergy module (semi-structured data), and clinical notes (unstructured data).

RESULTS

n total 8124 ICU admissions were included, with 542 (6.7%) ICU admissions experiencing AKI stage 2 or 3. The ICU physicians deemed 102 of these AKI cases (18.8%) to be drug-related. These DAKI cases were all documented in the clinical notes (100%), one in allergy module (1%) and none via diagnosis codes. The clinical notes required the highest time investment to analyze.

CONCLUSIONS

Drug-related causes comprise a substantial part of AKI in the ICU patients. However, current unstructured DAKI documentation practice via clinical notes hampers our ability to gain better insights about DAKI occurrence. Therefore, both automating DAKI identification from the clinical notes and increasing structured DAKI documentation should be encouraged.

Evaluation of the portability of computable phenotypes with natural language processing in the eMERGE network

The electronic Medical Records and Genomics (eMERGE) Network assessed the feasibility of deploying portable phenotype rule-based algorithms with natural language processing (NLP) components added to improve performance of existing algorithms using electronic health records (EHRs). Based on scientific merit and predicted difficulty, eMERGE selected six existing phenotypes to enhance with NLP. We assessed performance, portability, and ease of use. We summarized lessons learned by: (1) challenges; (2) best practices to address challenges based on existing evidence and/or eMERGE experience; and (3) opportunities for future research. Adding NLP resulted in improved, or the same, precision and/or recall for all but one algorithm. Portability, phenotyping workflow/process, and technology were major themes. With NLP, development and validation took longer. Besides portability of NLP technology and algorithm replicability, factors to ensure success include privacy protection, technical infrastructure setup, intellectual property agreement, and efficient communication. Workflow improvements can improve communication and reduce implementation time. NLP performance varied mainly due to clinical document heterogeneity; therefore, we suggest using semi-structured notes, comprehensive documentation, and customization options. NLP portability is possible with improved phenotype algorithm performance, but careful planning and architecture of the algorithms is essential to support local customizations.

Improving Methods of Identifying Anaphylaxis for Medical Product Safety Surveillance Using Natural Language Processing and Machine Learning

We sought to determine whether machine learning and natural language processing (NLP) applied to electronic medical records could improve performance of automated health-care claims-based algorithms to identify anaphylaxis events using data on 516 patients with outpatient, emergency department, or inpatient anaphylaxis diagnosis codes during 2015-2019 in 2 integrated health-care institutions in the Northwest United States. We used one site's manually reviewed gold-standard outcomes data for model development and the other's for external validation based on cross-validated area under the receiver operating characteristic curve (AUC), positive predictive value (PPV), and sensitivity. In the development site 154 (64%) of 239 potential events met adjudication criteria for anaphylaxis compared with 180 (65%) of 277 in the validation site. Logistic regression models using only structured claims data achieved a cross-validated AUC of 0.58 (95% CI: 0.54, 0.63). Machine learning improved cross-validated AUC to 0.62 (0.58, 0.66); incorporating NLP-derived covariates further increased cross-validated AUCs to 0.70 (0.66, 0.75) in development and 0.67 (0.63, 0.71) in external validation data. A classification threshold with cross-validated PPV of 79% and cross-validated sensitivity of 66% in development data had cross-validated PPV of 78% and cross-validated sensitivity of 56% in external data. Machine learning and NLP-derived data improved identification of validated anaphylaxis events.

Machine learning approaches for electronic health records phenotyping: a methodical review

OBJECTIVE

Accurate and rapid phenotyping is a prerequisite to leveraging electronic health records for biomedical research. While early phenotyping relied on rule-based algorithms curated by experts, machine learning (ML) approaches have emerged as an alternative to improve scalability across phenotypes and healthcare settings. This study evaluates ML-based phenotyping with respect to (1) the data sources used, (2) the phenotypes considered, (3) the methods applied, and (4) the reporting and evaluation methods used.

MATERIALS AND METHODS

We searched PubMed and Web of Science for articles published between 2018 and 2022. After screening 850 articles, we recorded 37 variables on 100 studies.

RESULTS

Most studies utilized data from a single institution and included information in clinical notes. Although chronic conditions were most commonly considered, ML also enabled the characterization of nuanced phenotypes such as social determinants of health. Supervised deep learning was the most popular ML paradigm, while semi-supervised and weakly supervised learning were applied to expedite algorithm development and unsupervised learning to facilitate phenotype discovery. ML approaches did not uniformly outperform rule-based algorithms, but deep learning offered a marginal improvement over traditional ML for many conditions.

DISCUSSION

Despite the progress in ML-based phenotyping, most articles focused on binary phenotypes and few articles evaluated external validity or used multi-institution data. Study settings were infrequently reported and analytic code was rarely released.

CONCLUSION

Continued research in ML-based phenotyping is warranted, with emphasis on characterizing nuanced phenotypes, establishing reporting and evaluation standards, and developing methods to accommodate misclassified phenotypes due to algorithm errors in downstream applications.

Efficient Evaluation of Prediction Rules in Semi-Supervised Settings under Stratified Sampling

In many contemporary applications, large amounts of unlabeled data are readily available while labeled examples are limited. There has been substantial interest in semi-supervised learning (SSL) which aims to leverage unlabeled data to improve estimation or prediction. However, current SSL literature focuses primarily on settings where labeled data is selected uniformly at random from the population of interest. Stratified sampling, while posing additional analytical challenges, is highly applicable to many real world problems. Moreover, no SSL methods currently exist for estimating the prediction performance of a fitted model when the labeled data is not selected uniformly at random. In this paper, we propose a two-step SSL procedure for evaluating a prediction rule derived from a working binary regression model based on the Brier score and overall misclassification rate under stratified sampling. In step I, we impute the missing labels via weighted regression with nonlinear basis functions to account for stratified sampling and to improve efficiency. In step II, we augment the initial imputations to ensure the consistency of the resulting estimators regardless of the specification of the prediction model or the imputation model. The final estimator is then obtained with the augmented imputations. We provide asymptotic theory and numerical studies illustrating that our proposals outperform their supervised counterparts in terms of efficiency gain. Our methods are motivated by electronic health record (EHR) research and validated with a real data analysis of an EHR-based study of diabetic neuropathy.

MixEHR-Guided: A guided multi-modal topic modeling approach for large-scale automatic phenotyping using the electronic health record

Electronic Health Records (EHRs) contain rich clinical data collected at the point of the care, and their increasing adoption offers exciting opportunities for clinical informatics, disease risk prediction, and personalized treatment recommendation. However, effective use of EHR data for research and clinical decision support is often hampered by a lack of reliable disease labels. To compile gold-standard labels, researchers often rely on clinical experts to develop rule-based phenotyping algorithms from billing codes and other surrogate features. This process is tedious and error-prone due to recall and observer biases in how codes and measures are selected, and some phenotypes are incompletely captured by a handful of surrogate features. To address this challenge, we present a novel automatic phenotyping model called MixEHR-Guided (MixEHR-G), a multimodal hierarchical Bayesian topic model that efficiently models the EHR generative process by identifying latent phenotype structure in the data. Unlike existing topic modeling algorithms wherein the inferred topics are not identifiable, MixEHR-G uses prior information from informative surrogate features to align topics with known phenotypes. We applied MixEHR-G to an openly-available EHR dataset of 38,597 intensive care patients (MIMIC-III) in Boston, USA and to administrative claims data for a population-based cohort (PopHR) of 1.3 million people in Quebec, Canada. Qualitatively, we demonstrate that MixEHR-G learns interpretable phenotypes and yields meaningful insights about phenotype similarities, comorbidities, and epidemiological associations. Quantitatively, MixEHR-G outperforms existing unsupervised phenotyping methods on a phenotype label annotation task, and it can accurately estimate relative phenotype prevalence functions without gold-standard phenotype information. Altogether, MixEHR-G is an important step towards building an interpretable and automated phenotyping system using EHR data.

Natural Language Processing to Improve Prediction of Incident Atrial Fibrillation Using Electronic Health Records

Background Models predicting atrial fibrillation (AF) risk, such as Cohorts for Heart and Aging Research in Genomic Epidemiology AF (CHARGE-AF), have not performed as well in electronic health records. Natural language processing (NLP) may improve models by using narrative electronic health record text. Methods and Results From a primary care network, we included patients aged ≥65 years with visits between 2003 and 2013 in development (n=32 960) and internal validation cohorts (n=13 992). An external validation cohort from a separate network from 2015 to 2020 included 39 051 patients. Model features were defined using electronic health record codified data and narrative data with NLP. We developed 2 models to predict 5-year AF incidence using (1) codified+NLP data and (2) codified data only and evaluated model performance. The analysis included 2839 incident AF cases in the development cohort and 1057 and 2226 cases in internal and external validation cohorts, respectively. The C-statistic was greater (P<0.001) in codified+NLP model (0.744 [95% CI, 0.735-0.753]) compared with codified-only (0.730 [95% CI, 0.720-0.739]) in the development cohort. In internal validation, the C-statistic of codified+NLP was modestly higher (0.735 [95% CI, 0.720-0.749]) compared with codified-only (0.729 [95% CI, 0.715-0.744]; P=0.06) and CHARGE-AF (0.717 [95% CI, 0.703-0.731]; P=0.002). Codified+NLP and codified-only were well calibrated, whereas CHARGE-AF underestimated AF risk. In external validation, the C-statistic of codified+NLP (0.750 [95% CI, 0.740-0.760]) remained higher (P<0.001) than codified-only (0.738 [95% CI, 0.727-0.748]) and CHARGE-AF (0.735 [95% CI, 0.725-0.746]). Conclusions Estimation of 5-year risk of AF can be modestly improved using NLP to incorporate narrative electronic health record data.

Scalable relevance ranking algorithm via semantic similarity assessment improves efficiency of medical chart review

OBJECTIVE

Accurately assigning phenotype information to individual patients via computational phenotyping using Electronic Health Records (EHRs) has been seen as the first step towards enabling EHRs for precision medicine research. Chart review labels annotated by clinical experts, also known as "gold standard" labels, are essential for the development and validation of computational phenotyping algorithms. However, given the complexity of EHR systems, the process of chart review is both labor intensive and time consuming. We propose a fully automated algorithm, referred to as pGUESS, to rank EHR notes according to their relevance to a given phenotype. By identifying the most relevant notes, pGUESS can greatly improve the efficiency and accuracy of chart reviews.

METHOD

pGUESS uses prior guided semantic similarity to measure the informativeness of a clinical note to a given phenotype. We first select candidate clinical concepts from a pool of comprehensive medical concepts using public knowledge sources and then derive the semantic embedding vector (SEV) for a reference article (SEV_ref) and each note (SEV_note). The algorithm scores the relevance of a note as the cosine similarity between SEV_note and SEV_ref.

RESULTS

The algorithm was validated against four sets of 200 notes that were manually annotated by clinical experts to assess their informativeness to one of three disease phenotypes. pGUESS algorithm substantially outperforms existing unsupervised approaches for classifying the relevance status with respect to both accuracy and scalability across phenotypes. Averaging over the three phenotypes, the rank correlation between the algorithm ranking and gold standard label was 0.64 for pGUESS, but only 0.47 and 0.35 for the next two best performing algorithms. pGUESS is also much more computationally scalable compared to existing algorithms.

CONCLUSION

pGUESS algorithm can substantially reduce the burden of chart review and holds potential in improving the efficiency and accuracy of human annotation.

Semi-supervised approach to event time annotation using longitudinal electronic health records

Large clinical datasets derived from insurance claims and electronic health record (EHR) systems are valuable sources for precision medicine research. These datasets can be used to develop models for personalized prediction of risk or treatment response. Efficiently deriving prediction models using real world data, however, faces practical and methodological challenges. Precise information on important clinical outcomes such as time to cancer progression are not readily available in these databases. The true clinical event times typically cannot be approximated well based on simple extracts of billing or procedure codes. Whereas, annotating event times manually is time and resource prohibitive. In this paper, we propose a two-step semi-supervised multi-modal automated time annotation (MATA) method leveraging multi-dimensional longitudinal EHR encounter records. In step I, we employ a functional principal component analysis approach to estimate the underlying intensity functions based on observed point processes from the unlabeled patients. In step II, we fit a penalized proportional odds model to the event time outcomes with features derived in step I in the labeled data where the non-parametric baseline function is approximated using B-splines. Under regularity conditions, the resulting estimator of the feature effect vector is shown as root-n consistent. We demonstrate the superiority of our approach relative to existing approaches through simulations and a real data example on annotating lung cancer recurrence in an EHR cohort of lung cancer patients from Veteran Health Administration.

A multi-ethnic polygenic risk score is associated with hypertension prevalence and progression throughout adulthood

In a multi-stage analysis of 52,436 individuals aged 17-90 across diverse cohorts and biobanks, we train, test, and evaluate a polygenic risk score (PRS) for hypertension risk and progression. The PRS is trained using genome-wide association studies (GWAS) for systolic, diastolic blood pressure, and hypertension, respectively. For each trait, PRS is selected by optimizing the coefficient of variation (CV) across estimated effect sizes from multiple potential PRS using the same GWAS, after which the 3 trait-specific PRSs are combined via an unweighted sum called "PRSsum", forming the HTN-PRS. The HTN-PRS is associated with both prevalent and incident hypertension at 4-6 years of follow up. This association is further confirmed in age-stratified analysis. In an independent biobank of 40,201 individuals, the HTN-PRS is confirmed to be predictive of increased risk for coronary artery disease, ischemic stroke, type 2 diabetes, and chronic kidney disease.

PhenoDEF: a corpus for annotating sentences with information of phenotype definitions in biomedical literature

BACKGROUND

Adverse events induced by drug-drug interactions are a major concern in the United States. Current research is moving toward using electronic health record (EHR) data, including for adverse drug events discovery. One of the first steps in EHR-based studies is to define a phenotype for establishing a cohort of patients. However, phenotype definitions are not readily available for all phenotypes. One of the first steps of developing automated text mining tools is building a corpus. Therefore, this study aimed to develop annotation guidelines and a gold standard corpus to facilitate building future automated approaches for mining phenotype definitions contained in the literature. Furthermore, our aim is to improve the understanding of how these published phenotype definitions are presented in the literature and how we annotate them for future text mining tasks.

RESULTS

Two annotators manually annotated the corpus on a sentence-level for the presence of evidence for phenotype definitions. Three major categories (inclusion, intermediate, and exclusion) with a total of ten dimensions were proposed characterizing major contextual patterns and cues for presenting phenotype definitions in published literature. The developed annotation guidelines were used to annotate the corpus that contained 3971 sentences: 1923 out of 3971 (48.4%) for the inclusion category, 1851 out of 3971 (46.6%) for the intermediate category, and 2273 out of 3971 (57.2%) for exclusion category. The highest number of annotated sentences was 1449 out of 3971 (36.5%) for the "Biomedical & Procedure" dimension. The lowest number of annotated sentences was 49 out of 3971 (1.2%) for "The use of NLP". The overall percent inter-annotator agreement was 97.8%. Percent and Kappa statistics also showed high inter-annotator agreement across all dimensions.

CONCLUSIONS

The corpus and annotation guidelines can serve as a foundational informatics approach for annotating and mining phenotype definitions in literature, and can be used later for text mining applications.

Development of a clinical polygenic risk score assay and reporting workflow

Implementation of polygenic risk scores (PRS) may improve disease prevention and management but poses several challenges: the construction of clinically valid assays, interpretation for individual patients, and the development of clinical workflows and resources to support their use in patient care. For the ongoing Veterans Affairs Genomic Medicine at Veterans Affairs (GenoVA) Study we developed a clinical genotype array-based assay for six published PRS. We used data from 36,423 Mass General Brigham Biobank participants and adjustment for population structure to replicate known PRS-disease associations and published PRS thresholds for a disease odds ratio (OR) of 2 (ranging from 1.75 (95% CI: 1.57-1.95) for type 2 diabetes to 2.38 (95% CI: 2.07-2.73) for breast cancer). After confirming the high performance and robustness of the pipeline for use as a clinical assay for individual patients, we analyzed the first 227 prospective samples from the GenoVA Study and found that the frequency of PRS corresponding to published OR > 2 ranged from 13/227 (5.7%) for colorectal cancer to 23/150 (15.3%) for prostate cancer. In addition to the PRS laboratory report, we developed physician- and patient-oriented informational materials to support decision-making about PRS results. Our work illustrates the generalizable development of a clinical PRS assay for multiple conditions and the technical, reporting and clinical workflow challenges for implementing PRS information in the clinic.

Metabolomic profiling reveals extensive adrenal suppression due to inhaled corticosteroid therapy in asthma

The application of large-scale metabolomic profiling provides new opportunities for realizing the potential of omics-based precision medicine for asthma. By leveraging data from over 14,000 individuals in four distinct cohorts, this study identifies and independently replicates 17 steroid metabolites whose levels were significantly reduced in individuals with prevalent asthma. Although steroid levels were reduced among all asthma cases regardless of medication use, the largest reductions were associated with inhaled corticosteroid (ICS) treatment, as confirmed in a 4-year low-dose ICS clinical trial. Effects of ICS treatment on steroid levels were dose dependent; however, significant reductions also occurred with low-dose ICS treatment. Using information from electronic medical records, we found that cortisol levels were substantially reduced throughout the entire 24-hour daily period in patients with asthma who were treated with ICS compared to those who were untreated and to patients without asthma. Moreover, patients with asthma who were treated with ICS showed significant increases in fatigue and anemia as compared to those without ICS treatment. Adrenal suppression in patients with asthma treated with ICS might, therefore, represent a larger public health problem than previously recognized. Regular cortisol monitoring of patients with asthma treated with ICS is needed to provide the optimal balance between minimizing adverse effects of adrenal suppression while capitalizing on the established benefits of ICS treatment.

Validation of a computational phenotype for finding patients eligible for genetic testing for pathogenic PTEN variants across three centers

BACKGROUND

Computational phenotypes are most often combinations of patient billing codes that are highly predictive of disease using electronic health records (EHR). In the case of rare diseases that can only be diagnosed by genetic testing, computational phenotypes identify patient cohorts for genetic testing and possible diagnosis. This article details the validation of a computational phenotype for PTEN hamartoma tumor syndrome (PHTS) against the EHR of patients at three collaborating clinical research centers: Boston Children's Hospital, Children's National Hospital, and the University of Washington.

METHODS

A combination of billing codes from the International Classification of Diseases versions 9 and 10 (ICD-9 and ICD-10) for diagnostic criteria postulated by a research team at Cleveland Clinic was used to identify patient cohorts for genetic testing from the clinical data warehouses at the three research centers. Subsequently, the EHR-including billing codes, clinical notes, and genetic reports-of these patients were reviewed by clinical experts to identify patients with PHTS.

RESULTS

The PTEN genetic testing yield of the computational phenotype, the number of patients who needed to be genetically tested for incidence of pathogenic PTEN gene variants, ranged from 82 to 94% at the three centers.

CONCLUSIONS

Computational phenotypes have the potential to enable the timely and accurate diagnosis of rare genetic diseases such as PHTS by identifying patient cohorts for genetic sequencing and testing.

Recessive Genome-Wide Meta-analysis Illuminates Genetic Architecture of Type 2 Diabetes

Most genome-wide association studies (GWAS) of complex traits are performed using models with additive allelic effects. Hundreds of loci associated with type 2 diabetes have been identified using this approach. Additive models, however, can miss loci with recessive effects, thereby leaving potentially important genes undiscovered. We conducted the largest GWAS meta-analysis using a recessive model for type 2 diabetes. Our discovery sample included 33,139 case subjects and 279,507 control subjects from 7 European-ancestry cohorts, including the UK Biobank. We identified 51 loci associated with type 2 diabetes, including five variants undetected by prior additive analyses. Two of the five variants had minor allele frequency of <5% and were each associated with more than a doubled risk in homozygous carriers. Using two additional cohorts, FinnGen and a Danish cohort, we replicated three of the variants, including one of the low-frequency variants, rs115018790, which had an odds ratio in homozygous carriers of 2.56 (95% CI 2.05-3.19; P = 1 × 10-16) and a stronger effect in men than in women (for interaction, P = 7 × 10-7). The signal was associated with multiple diabetes-related traits, with homozygous carriers showing a 10% decrease in LDL cholesterol and a 20% increase in triglycerides; colocalization analysis linked this signal to reduced expression of the nearby PELO gene. These results demonstrate that recessive models, when compared with GWAS using the additive approach, can identify novel loci, including large-effect variants with pathophysiological consequences relevant to type 2 diabetes.

Prior Adaptive Semi-supervised Learning with Application to EHR Phenotyping

Electronic Health Record (EHR) data, a rich source for biomedical research, have been successfully used to gain novel insight into a wide range of diseases. Despite its potential, EHR is currently underutilized for discovery research due to its major limitation in the lack of precise phenotype information. To overcome such difficulties, recent efforts have been devoted to developing supervised algorithms to accurately predict phenotypes based on relatively small training datasets with gold standard labels extracted via chart review. However, supervised methods typically require a sizable training set to yield generalizable algorithms, especially when the number of candidate features, p , is large. In this paper, we propose a semi-supervised (SS) EHR phenotyping method that borrows information from both a small, labeled dataset (where both the label Y and the feature set X are observed) and a much larger, weakly-labeled dataset in which the feature set X is accompanied only by a surrogate label S that is available to all patients. Under a working prior assumption that S is related to X only through Y and allowing it to hold approximately, we propose a prior adaptive semi-supervised (PASS) estimator that incorporates the prior knowledge by shrinking the estimator towards a direction derived under the prior. We derive asymptotic theory for the proposed estimator and justify its efficiency and robustness to prior information of poor quality. We also demonstrate its superiority over existing estimators under various scenarios via simulation studies and on three real-world EHR phenotyping studies at a large tertiary hospital.

SAT: a Surrogate-Assisted Two-wave case boosting sampling method, with application to EHR-based association studies

OBJECTIVES

Electronic health records (EHRs) enable investigation of the association between phenotypes and risk factors. However, studies solely relying on potentially error-prone EHR-derived phenotypes (ie, surrogates) are subject to bias. Analyses of low prevalence phenotypes may also suffer from poor efficiency. Existing methods typically focus on one of these issues but seldom address both. This study aims to simultaneously address both issues by developing new sampling methods to select an optimal subsample to collect gold standard phenotypes for improving the accuracy of association estimation.

MATERIALS AND METHODS

We develop a surrogate-assisted two-wave (SAT) sampling method, where a surrogate-guided sampling (SGS) procedure and a modified optimal subsampling procedure motivated from A-optimality criterion (OSMAC) are employed sequentially, to select a subsample for outcome validation through manual chart review subject to budget constraints. A model is then fitted based on the subsample with the true phenotypes. Simulation studies and an application to an EHR dataset of breast cancer survivors are conducted to demonstrate the effectiveness of SAT.

RESULTS

We found that the subsample selected with the proposed method contains informative observations that effectively reduce the mean squared error of the resultant estimator of the association.

CONCLUSIONS

The proposed approach can handle the problem brought by the rarity of cases and misclassification of the surrogate in phenotype-absent EHR-based association studies. With a well-behaved surrogate, SAT successfully boosts the case prevalence in the subsample and improves the efficiency of estimation.

Machine learning can identify patients at risk of hyperparathyroidism without known calcium and intact parathyroid hormone

BACKGROUND

To prove the concept of diagnosing primary hyperparathyroidism (pHPT) without calcium and parathyroid hormone (PTH) values and identifying potential risk factors for pHPT.

METHODS

Data were extracted from the clinical data warehouse (CDW) at the University of Arkansas for Medical Sciences (UAMS) Epic EHR (2014-2019).

RESULTS

1737 patients with over 185 000 rows of clinical data were provided in a relational structure and processed/flattened to facilitate modeling. Phenotype elements were identified for pHPT without advance knowledge of calcium and PTH levels. The area under the curve (AUC) for the prediction of pHPT using our model was 0.86 with sensitivity and specificity of 0.8953 and 0.6686, respectively, using a 0.45 probability threshold.

CONCLUSION

Primary hyperparathyroidism was predicted from a dataset excluding calcium and PTH data with 86% accuracy. This approach needs to be validated/refined on larger samples of data and plans are in place to do this with other regional/national datasets.

Pharmaco-Metabolomics of Inhaled Corticosteroid Response in Individuals with Asthma

Metabolomic indicators of asthma treatment responses have yet to be identified. In this study, we aimed to uncover plasma metabolomic profiles associated with asthma exacerbations while on inhaled corticosteroid (ICS) treatment. We determined whether these profiles change with age from adolescence to adulthood. We utilized data from 170 individuals with asthma on ICS from the Mass General Brigham Biobank to identify plasma metabolites associated with asthma exacerbations while on ICS and examined potential effect modification of metabolite-exacerbation associations by age. We used liquid chromatography-high-resolution mass spectrometry-based metabolomic profiling. Sex-stratified analyses were also performed for the significant associations. The age range of the participating individuals was 13-43 years with a mean age of 33.5 years. Of the 783 endogenous metabolites tested, eight demonstrated significant associations with exacerbation after correction for multiple comparisons and adjusting for potential confounders (Bonferroni p value < 6.2 × 10^-4). Potential effect modification by sex was detected for fatty acid metabolites, with males showing a greater reduction in their metabolite levels with ICS exacerbation. Thirty-eight metabolites showed suggestive interactions with age on exacerbation (nominal p-value < 0.05). Our findings demonstrate that plasma metabolomic profiles differ for individuals who experience asthma exacerbations while on ICS. The differentiating metabolites may serve as biomarkers of ICS response and may highlight metabolic pathways underlying ICS response variability.

High-throughput phenotyping with temporal sequences

OBJECTIVE

High-throughput electronic phenotyping algorithms can accelerate translational research using data from electronic health record (EHR) systems. The temporal information buried in EHRs is often underutilized in developing computational phenotypic definitions. This study aims to develop a high-throughput phenotyping method, leveraging temporal sequential patterns from EHRs.

MATERIALS AND METHODS

We develop a representation mining algorithm to extract 5 classes of representations from EHR diagnosis and medication records: the aggregated vector of the records (aggregated vector representation), the standard sequential patterns (sequential pattern mining), the transitive sequential patterns (transitive sequential pattern mining), and 2 hybrid classes. Using EHR data on 10 phenotypes from the Mass General Brigham Biobank, we train and validate phenotyping algorithms.

RESULTS

Phenotyping with temporal sequences resulted in a superior classification performance across all 10 phenotypes compared with the standard representations in electronic phenotyping. The high-throughput algorithm's classification performance was superior or similar to the performance of previously published electronic phenotyping algorithms. We characterize and evaluate the top transitive sequences of diagnosis records paired with the records of risk factors, symptoms, complications, medications, or vaccinations.

DISCUSSION

The proposed high-throughput phenotyping approach enables seamless discovery of sequential record combinations that may be difficult to assume from raw EHR data. Transitive sequences offer more accurate characterization of the phenotype, compared with its individual components, and reflect the actual lived experiences of the patients with that particular disease.

CONCLUSION

Sequential data representations provide a precise mechanism for incorporating raw EHR records into downstream machine learning. Our approach starts with user interpretability and works backward to the technology.

Individual Data Protected Integrative Regression Analysis of High-Dimensional Heterogeneous Data

Evidence-based decision making often relies on meta-analyzing multiple studies, which enables more precise estimation and investigation of generalizability. Integrative analysis of multiple heterogeneous studies is, however, highly challenging in the ultra high-dimensional setting. The challenge is even more pronounced when the individual-level data cannot be shared across studies, known as DataSHIELD contraint. Under sparse regression models that are assumed to be similar yet not identical across studies, we propose in this paper a novel integrative estimation procedure for data-Shielding High-dimensional Integrative Regression (SHIR). SHIR protects individual data through summary-statistics-based integrating procedure, accommodates between-study heterogeneity in both the covariate distribution and model parameters, and attains consistent variable selection. Theoretically, SHIR is statistically more efficient than the existing distributed approaches that integrate debiased LASSO estimators from the local sites. Furthermore, the estimation error incurred by aggregating derived data is negligible compared to the statistical minimax rate and SHIR is shown to be asymptotically equivalent in estimation to the ideal estimator obtained by sharing all data. The finite-sample performance of our method is studied and compared with existing approaches via extensive simulation settings. We further illustrate the utility of SHIR to derive phenotyping algorithms for coronary artery disease using electronic health records data from multiple chronic disease cohorts.

CQL4NLP: Development and Integration of FHIR NLP Extensions in Clinical Quality Language for EHR-driven Phenotyping

Lack of standardized representation of natural language processing (NLP) components in phenotyping algorithms hinders portability of the phenotyping algorithms and their execution in a high-throughput and reproducible manner. The objective of the study is to develop and evaluate a standard-driven approach - CQL4NLP - that integrates a collection of NLP extensions represented in the HL7 Fast Healthcare Interoperability Resources (FHIR) standard into the clinical quality language (CQL). A minimal NLP data model with 11 NLP-specific data elements was created, including six FHIR NLP extensions. All 11 data elements were identified from their usage in real-world phenotyping algorithms. An NLP ruleset generation mechanism was integrated into the NLP2FHIR pipeline and the NLP rulesets enabled comparable performance for a case study with the identification of obesity comorbidities. The NLP ruleset generation mechanism created a reproducible process for defining the NLP components of a phenotyping algorithm and its execution.

Comparative effectiveness of medical concept embedding for feature engineering in phenotyping

OBJECTIVE

Feature engineering is a major bottleneck in phenotyping. Properly learned medical concept embeddings (MCEs) capture the semantics of medical concepts, thus are useful for retrieving relevant medical features in phenotyping tasks. We compared the effectiveness of MCEs learned from knowledge graphs and electronic healthcare records (EHR) data in retrieving relevant medical features for phenotyping tasks.

MATERIALS AND METHODS

We implemented 5 embedding methods including node2vec, singular value decomposition (SVD), LINE, skip-gram, and GloVe with 2 data sources: (1) knowledge graphs obtained from the observational medical outcomes partnership (OMOP) common data model; and (2) patient-level data obtained from the OMOP compatible electronic health records (EHR) from Columbia University Irving Medical Center (CUIMC). We used phenotypes with their relevant concepts developed and validated by the electronic medical records and genomics (eMERGE) network to evaluate the performance of learned MCEs in retrieving phenotype-relevant concepts. Hits@k% in retrieving phenotype-relevant concepts based on a single and multiple seed concept(s) was used to evaluate MCEs.

RESULTS

Among all MCEs, MCEs learned by using node2vec with knowledge graphs showed the best performance. Of MCEs based on knowledge graphs and EHR data, MCEs learned by using node2vec with knowledge graphs and MCEs learned by using GloVe with EHR data outperforms other MCEs, respectively.

CONCLUSION

MCE enables scalable feature engineering tasks, thereby facilitating phenotyping. Based on current phenotyping practices, MCEs learned by using knowledge graphs constructed by hierarchical relationships among medical concepts outperformed MCEs learned by using EHR data.

Health-adjusted life expectancy (HALE) in Chongqing, China, 2017: An artificial intelligence and big data method estimating the burden of disease at city level

BACKGROUND

A universally applicable approach that provides standard HALE measurements for different regions has yet to be developed because of the difficulties of health information collection. In this study, we developed a natural language processing (NLP) based HALE estimation approach by using individual-level electronic medical records (EMRs), which made it possible to calculate HALE timely in different temporal or spatial granularities.

METHODS

We performed diagnostic concept extraction and normalisation on 13•99 million EMRs with NLP to estimate the prevalence of 254 diseases in WHO Global Burden of Disease Study (GBD). Then, we calculated HALE in Chongqing, 2017, by using the life table technique and Sullivan's method, and analysed the contribution of diseases to the expected years "lost" due to disability (DLE).

FINDINGS

Our method identified a life expectancy at birth (LE0) of 77•9 years and health-adjusted life expectancy at birth (HALE0) of 71•7 years for the general Chongqing population of 2017. In particular, the male LE0 and HALE0 were 76•3 years and 68•9 years, respectively, while the female LE0 and HALE0 were 80•0 years and 74•4 years, respectively. Cerebrovascular diseases, cancers, and injuries were the top three deterioration factors, which reduced HALE by 2•67, 2•15, and 1•19 years, respectively.

INTERPRETATION

The results demonstrated the feasibility and effectiveness of EMRs-based HALE estimation. Moreover, the method allowed for a potentially transferable framework that facilitated a more convenient comparison of cross-sectional and longitudinal studies on HALE between regions. In summary, this study provided insightful solutions to the global ageing and health problems that the world is facing.

FUNDING

National Key R and D Program of China (2018YFC2000400).

sureLDA: A multidisease automated phenotyping method for the electronic health record

OBJECTIVE

A major bottleneck hindering utilization of electronic health record data for translational research is the lack of precise phenotype labels. Chart review as well as rule-based and supervised phenotyping approaches require laborious expert input, hampering applicability to studies that require many phenotypes to be defined and labeled de novo. Though International Classification of Diseases codes are often used as surrogates for true labels in this setting, these sometimes suffer from poor specificity. We propose a fully automated topic modeling algorithm to simultaneously annotate multiple phenotypes.

MATERIALS AND METHODS

Surrogate-guided ensemble latent Dirichlet allocation (sureLDA) is a label-free multidimensional phenotyping method. It first uses the PheNorm algorithm to initialize probabilities based on 2 surrogate features for each target phenotype, and then leverages these probabilities to constrain the LDA topic model to generate phenotype-specific topics. Finally, it combines phenotype-feature counts with surrogates via clustering ensemble to yield final phenotype probabilities.

RESULTS

sureLDA achieves reliably high accuracy and precision across a range of simulated and real-world phenotypes. Its performance is robust to phenotype prevalence and relative informativeness of surogate vs nonsurrogate features. It also exhibits powerful feature selection properties.

DISCUSSION

sureLDA combines attractive properties of PheNorm and LDA to achieve high accuracy and precision robust to diverse phenotype characteristics. It offers particular improvement for phenotypes insufficiently captured by a few surrogate features. Moreover, sureLDA's feature selection ability enables it to handle high feature dimensions and produce interpretable computational phenotypes.

CONCLUSIONS

sureLDA is well suited toward large-scale electronic health record phenotyping for highly multiphenotype applications such as phenome-wide association studies .

Generative transfer learning for measuring plausibility of EHR diagnosis records

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.

Integration of genetic and clinical information to improve imputation of data missing from electronic health records

OBJECTIVE

Clinical data of patients' measurements and treatment history stored in electronic health record (EHR) systems are starting to be mined for better treatment options and disease associations. A primary challenge associated with utilizing EHR data is the considerable amount of missing data. Failure to address this issue can introduce significant bias in EHR-based research. Currently, imputation methods rely on correlations among the structured phenotype variables in the EHR. However, genetic studies have shown that many EHR-based phenotypes have a heritable component, suggesting that measured genetic variants might be useful for imputing missing data. In this article, we developed a computational model that incorporates patients' genetic information to perform EHR data imputation.

MATERIALS AND METHODS

We used the individual single nucleotide polymorphism's association with phenotype variables in the EHR as input to construct a genetic risk score that quantifies the genetic contribution to the phenotype. Multiple approaches to constructing the genetic risk score were evaluated for optimal performance. The genetic score, along with phenotype correlation, is then used as a predictor to impute the missing values.

RESULTS

To demonstrate the method performance, we applied our model to impute missing cardiovascular related measurements including low-density lipoprotein, heart failure, and aortic aneurysm disease in the electronic Medical Records and Genomics data. The integration method improved imputation's area-under-the-curve for binary phenotypes and decreased root-mean-square error for continuous phenotypes.

CONCLUSION

Compared with standard imputation approaches, incorporating genetic information offers a novel approach that can utilize more of the EHR data for better performance in missing data imputation.

High-throughput multimodal automated phenotyping (MAP) with application to PheWAS

OBJECTIVE

Electronic health records linked with biorepositories are a powerful platform for translational studies. A major bottleneck exists in the ability to phenotype patients accurately and efficiently. The objective of this study was to develop an automated high-throughput phenotyping method integrating International Classification of Diseases (ICD) codes and narrative data extracted using natural language processing (NLP).

MATERIALS AND METHODS

We developed a mapping method for automatically identifying relevant ICD and NLP concepts for a specific phenotype leveraging the Unified Medical Language System. Along with health care utilization, aggregated ICD and NLP counts were jointly analyzed by fitting an ensemble of latent mixture models. The multimodal automated phenotyping (MAP) algorithm yields a predicted probability of phenotype for each patient and a threshold for classifying participants with phenotype yes/no. The algorithm was validated using labeled data for 16 phenotypes from a biorepository and further tested in an independent cohort phenome-wide association studies (PheWAS) for 2 single nucleotide polymorphisms with known associations.

RESULTS

The MAP algorithm achieved higher or similar AUC and F-scores compared to the ICD code across all 16 phenotypes. The features assembled via the automated approach had comparable accuracy to those assembled via manual curation (AUCMAP 0.943, AUCmanual 0.941). The PheWAS results suggest that the MAP approach detected previously validated associations with higher power when compared to the standard PheWAS method based on ICD codes.

CONCLUSION

The MAP approach increased the accuracy of phenotype definition while maintaining scalability, thereby facilitating use in studies requiring large-scale phenotyping, such as PheWAS.