Dataset Information

Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records.

ABSTRACT: Machine learning has become ubiquitous and a key technology on mining electronic health records (EHRs) for facilitating clinical research and practice. Unsupervised machine learning, as opposed to supervised learning, has shown promise in identifying novel patterns and relations from EHRs without using human created labels. In this paper, we investigate the application of unsupervised machine learning models in discovering latent disease clusters and patient subgroups based on EHRs. We utilized Latent Dirichlet Allocation (LDA), a generative probabilistic model, and proposed a novel model named Poisson Dirichlet Model (PDM), which extends the LDA approach using a Poisson distribution to model patients' disease diagnoses and to alleviate age and sex factors by considering both observed and expected observations. In the empirical experiments, we evaluated LDA and PDM on three patient cohorts, namely Osteoporosis, Delirium/Dementia, and Chronic Obstructive Pulmonary Disease (COPD)/Bronchiectasis Cohorts, with their EHR data retrieved from the Rochester Epidemiology Project (REP) medical records linkage system, for the discovery of latent disease clusters and patient subgroups. We compared the effectiveness of LDA and PDM in identifying disease clusters through the visualization of disease representations. We tested the performance of LDA and PDM in differentiating patient subgroups through survival analysis, as well as statistical analysis of demographics and Elixhauser Comorbidity Index (ECI) scores in those subgroups. The experimental results show that the proposed PDM could effectively identify distinguished disease clusters based on the latent patterns hidden in the EHR data by alleviating the impact of age and sex, and that LDA could stratify patients into differentiable subgroups with larger p-values than PDM. However, those subgroups identified by LDA are highly associated with patients' age and sex. The subgroups discovered by PDM might imply the underlying patterns of diseases of greater interest in epidemiology research due to the alleviation of age and sex. Both unsupervised machine learning approaches could be leveraged to discover patient subgroups using EHRs but with different foci.

SUBMITTER: Wang Y

PROVIDER: S-EPMC7028517 | biostudies-literature | 2020 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records.

Wang Yanshan Y Zhao Yiqing Y Therneau Terry M TM Atkinson Elizabeth J EJ Tafti Ahmad P AP Zhang Nan N Amin Shreyasee S Limper Andrew H AH Khosla Sundeep S Liu Hongfang H

Journal of biomedical informatics 20191228

Machine learning has become ubiquitous and a key technology on mining electronic health records (EHRs) for facilitating clinical research and practice. Unsupervised machine learning, as opposed to supervised learning, has shown promise in identifying novel patterns and relations from EHRs without using human created labels. In this paper, we investigate the application of unsupervised machine learning models in discovering latent disease clusters and patient subgroups based on EHRs. We utilized ...[more]

PMID: 31891765

Similar Datasets

Project description:BackgroundAlzheimer's disease (AD) is a highly heterogeneous disease with diverse trajectories and outcomes observed in clinical populations. Understanding this heterogeneity can enable better treatment, prognosis and disease management. Studies to date have mainly used imaging or cognition data and have been limited in terms of data breadth and sample size. Here we examine the clinical heterogeneity of Alzheimer's disease patients using electronic health records (EHR) to identify and characterise disease subgroups using multiple clustering methods, identifying clusters which are clinically actionable.MethodsWe identified AD patients in primary care EHR from the Clinical Practice Research Datalink (CPRD) using a previously validated rule-based phenotyping algorithm. We extracted and included a range of comorbidities, symptoms and demographic features as patient features. We evaluated four different clustering methods (k-means, kernel k-means, affinity propagation and latent class analysis) to cluster Alzheimer's disease patients. We compared clusters on clinically relevant outcomes and evaluated each method using measures of cluster structure, stability, efficiency of outcome prediction and replicability in external data sets.ResultsWe identified 7,913 AD patients, with a mean age of 82 and 66.2% female. We included 21 features in our analysis. We observed 5, 2, 5 and 6 clusters in k-means, kernel k-means, affinity propagation and latent class analysis respectively. K-means was found to produce the most consistent results based on four evaluative measures. We discovered a consistent cluster found in three of the four methods composed of predominantly female, younger disease onset (43% between ages 42-73) diagnosed with depression and anxiety, with a quicker rate of progression compared to the average across other clusters.ConclusionEach clustering approach produced substantially different clusters and K-Means performed the best out of the four methods based on the four evaluative criteria. However, the consistent appearance of one particular cluster across three of the four methods potentially suggests the presence of a distinct disease subtype that merits further exploration. Our study underlines the variability of the results obtained from different clustering approaches and the importance of systematically evaluating different approaches for identifying disease subtypes in complex EHR.

Project description:BackgroundData mining of electronic health records (EHRs) has a huge potential for improving clinical decision support and to help healthcare deliver precision medicine. Unfortunately, the rule-based and machine learning-based approaches used for natural language processing (NLP) in healthcare today all struggle with various shortcomings related to performance, efficiency, or transparency.MethodsIn this paper, we address these issues by presenting a novel method for NLP that implements unsupervised learning of word embeddings, semi-supervised learning for simplified and accelerated clinical vocabulary and concept building, and deterministic rules for fine-grained control of information extraction. The clinical language is automatically learnt, and vocabulary, concepts, and rules supporting a variety of NLP downstream tasks can further be built with only minimal manual feature engineering and tagging required from clinical experts. Together, these steps create an open processing pipeline that gradually refines the data in a transparent way, which greatly improves the interpretable nature of our method. Data transformations are thus made transparent and predictions interpretable, which is imperative for healthcare. The combined method also has other advantages, like potentially being language independent, demanding few domain resources for maintenance, and able to cover misspellings, abbreviations, and acronyms. To test and evaluate the combined method, we have developed a clinical decision support system (CDSS) named Information System for Clinical Concept Searching (ICCS) that implements the method for clinical concept tagging, extraction, and classification.ResultsIn empirical studies the method shows high performance (recall 92.6%, precision 88.8%, F-measure 90.7%), and has demonstrated its value to clinical practice. Here we employ a real-life EHR-derived dataset to evaluate the method's performance on the task of classification (i.e., detecting patient allergies) against a range of common supervised learning algorithms. The combined method achieves state-of-the-art performance compared to the alternative methods we evaluate. We also perform a qualitative analysis of common word embedding methods on the task of word similarity to examine their potential for supporting automatic feature engineering for clinical NLP tasks.ConclusionsBased on the promising results, we suggest more research should be aimed at exploiting the inherent synergies between unsupervised, supervised, and rule-based paradigms for clinical NLP.

Project description:BackgroundNowadays, patients with chronic diseases such as diabetes and hypertension have reached alarming numbers worldwide. These diseases increase the risk of developing acute complications and involve a substantial economic burden and demand for health resources. The widespread adoption of Electronic Health Records (EHRs) is opening great opportunities for supporting decision-making. Nevertheless, data extracted from EHRs are complex (heterogeneous, high-dimensional and usually noisy), hampering the knowledge extraction with conventional approaches.MethodsWe propose the use of the Denoising Autoencoder (DAE), a Machine Learning (ML) technique allowing to transform high-dimensional data into latent representations (LRs), thus addressing the main challenges with clinical data. We explore in this work how the combination of LRs with a visualization method can be used to map the patient data in a two-dimensional space, gaining knowledge about the distribution of patients with different chronic conditions. Furthermore, this representation can be also used to characterize the patient's health status evolution, which is of paramount importance in the clinical setting.ResultsTo obtain clinical LRs, we considered real-world data extracted from EHRs linked to the University Hospital of Fuenlabrada in Spain. Experimental results showed the great potential of DAEs to identify patients with clinical patterns linked to hypertension, diabetes and multimorbidity. The procedure allowed us to find patients with the same main chronic disease but different clinical characteristics. Thus, we identified two kinds of diabetic patients with differences in their drug therapy (insulin and non-insulin dependant), and also a group of women affected by hypertension and gestational diabetes. We also present a proof of concept for mapping the health status evolution of synthetic patients when considering the most significant diagnoses and drugs associated with chronic patients.ConclusionOur results highlighted the value of ML techniques to extract clinical knowledge, supporting the identification of patients with certain chronic conditions. Furthermore, the patient's health status progression on the two-dimensional space might be used as a tool for clinicians aiming to characterize health conditions and identify their more relevant clinical codes.

Project description:BackgroundThe main goal of this study is to explore the use of features representing patient-level electronic health record (EHR) data, generated by the unsupervised deep learning algorithm autoencoder, in predictive modeling. Since autoencoder features are unsupervised, this paper focuses on their general lower-dimensional representation of EHR information in a wide variety of predictive tasks.MethodsWe compare the model with autoencoder features to traditional models: logistic model with least absolute shrinkage and selection operator (LASSO) and Random Forest algorithm. In addition, we include a predictive model using a small subset of response-specific variables (Simple Reg) and a model combining these variables with features from autoencoder (Enhanced Reg). We performed the study first on simulated data that mimics real world EHR data and then on actual EHR data from eight Advocate hospitals.ResultsOn simulated data with incorrect categories and missing data, the precision for autoencoder is 24.16% when fixing recall at 0.7, which is higher than Random Forest (23.61%) and lower than LASSO (25.32%). The precision is 20.92% in Simple Reg and improves to 24.89% in Enhanced Reg. When using real EHR data to predict the 30-day readmission rate, the precision of autoencoder is 19.04%, which again is higher than Random Forest (18.48%) and lower than LASSO (19.70%). The precisions for Simple Reg and Enhanced Reg are 18.70 and 19.69% respectively. That is, Enhanced Reg can have competitive prediction performance compared to LASSO. In addition, results show that Enhanced Reg usually relies on fewer features under the setting of simulations of this paper.ConclusionsWe conclude that autoencoder can create useful features representing the entire space of EHR data and which are applicable to a wide array of predictive tasks. Together with important response-specific predictors, we can derive efficient and robust predictive models with less labor in data extraction and model training.

Project description:Background Heart failure is currently divided into three main forms, HFrEF, HFpEF, and HFmrEF, but its etiology is diverse and highly heterogeneous. Many studies reported a variety of novel subgroups in heart failure patients, with unsupervised machine learning methods. The aim of this scoping review is to provide insights into how these techniques can diagnose and manage HF faster and better, thus providing direction for future research and facilitating its routine use in clinical practice. Methods The review was performed following PRISMA-SCR guideline. We searched the PubMed database for eligible publications. Studies were included if they defined new subgroups in HF patients using clustering analysis methods, and excluded if they are (1) Reviews, commentary, or editorials, (2) Studies not about defining new sub-types, or (3) Studies not using unsupervised algorithms. All study screening and data extraction were conducted independently by two investigators and narrative integration of data extracted from included studies was performed. Results Of the 498 studies identified, 47 were included in the analysis. Most studies (61.7%) were published in 2020 and later. The largest number of studies (46.8%) coming from the United States, and most of the studies were authored and included in the same country. The most commonly used machine learning method was hierarchical cluster analysis (46.8%), the most commonly used cluster variable type was comorbidity (61.7%), and the least used cluster variable type was genomics (12.8%). Most of the studies used data sets of less than 500 patients (48.9%), and the sample size had negative correlation with the number of clustering variables. The majority of studies (85.1%) assessed the association between cluster grouping and at least one outcomes, with death and hospitalization being the most commonly used outcome measures. Conclusion This scoping review provides an overview of recent studies proposing novel HF subgroups based on clustering analysis. Differences were found in study design, study population, clustering methods and variables, and outcomes of interests, and we provided insights into how these studies were conducted and identify the knowledge gaps to guide future research.

Project description:BackgroundHealthcare professionals are required to access, interpret and generate patient data in the digital environment, and use this information to deliver and optimise patient care. Healthcare students are rarely exposed to the technology, or given the opportunity to use this during their training, which can impact on the digital competence of the graduating workforce. In this study we set out to develop and define domains of competence and associated learning outcomes needed by healthcare graduates to commence working in a digital healthcare environment.MethodA National Working Group was established in the UK to integrate Electronic Patient Records (EPRs) into undergraduate education for healthcare students studying medicine, pharmacy, nursing and midwifery. The working group, comprising 12 academic institutions and representatives from NHS England, NHS Digital and EPR system providers, met to discuss and document key learning outcomes required for using EPRs in the healthcare environment. Outcomes were grouped into six key domains and refined by the group prior to external review by experts working in medical education or with EPRs.ResultsSix key domains of competence and associated learning outcomes were identified and defined. External expert review provided iterative refinement and amendment. The agreed domains were: 1) Digital Health: work as a practitioner in the digital healthcare environment; 2) Accessing Data: access and interpret patient data to inform clinical decision-making; 3) Communication: communicate effectively with healthcare professionals and patients in the digital environment; 4) Generating data: generate data for and about patients within the EPR; 5) Multidisciplinary working: work with healthcare professionals with and alongside EPRs; and 6) Monitoring and audit: monitor and improve the quality and safety of healthcare.ConclusionThe six domains of competence and associated learning outcomes can be used by academics to guide the integration of EPRs into undergraduate healthcare programmes. This is key to ensuring that the future healthcare workforce can work with and alongside EPRs.

Project description:Objective: To explore the efficacy of anticoagulation in improving outcomes and safety of Coronavirus disease 2019 (COVID-19) patients in subgroups identified by clinical-based stratification and unsupervised machine learning. Methods: This single-center retrospective cohort study unselectively reviewed 2,272 patients with COVID-19 admitted to the Tongji Hospital between Jan 25 and Mar 23, 2020. The association between AC treatment and outcomes was investigated in the propensity score (PS) matched cohort and the full cohort by inverse probability of treatment weighting (IPTW) analysis. Subgroup analysis, identified by clinical-based stratification or unsupervised machine learning, was used to identify sub-phenotypes with meaningful clinical features and the target patients benefiting most from AC. Results: AC treatment was associated with lower in-hospital death risk either in the PS matched cohort or by IPTW analysis in the full cohort. A higher incidence of clinically relevant non-major bleeding (CRNMB) was observed in the AC group, but not major bleeding. Clinical subgroup analysis showed that, at admission, severe cases of COVID-19 clinical classification, mild acute respiratory distress syndrome (ARDS) cases, and patients with a D-dimer level ≥0.5 μg/mL, may benefit from AC. During the hospital stay, critical cases and severe ARDS cases may benefit from AC. Unsupervised machine learning analysis established a four-class clustering model. Clusters 1 and 2 were non-critical cases and might not benefit from AC, while clusters 3 and 4 were critical patients. Patients in cluster 3 might benefit from AC with no increase in bleeding events. While patients in cluster 4, who were characterized by multiple organ dysfunction (neurologic, circulation, coagulation, kidney and liver dysfunction) and elevated inflammation biomarkers, did not benefit from AC. Conclusions: AC treatment was associated with lower in-hospital death risk, especially in critically ill COVID-19 patients. Unsupervised learning analysis revealed that the most critically ill patients with multiple organ dysfunction and excessive inflammation might not benefit from AC. More attention should be paid to bleeding events (especially CRNMB) when using AC.

Dataset Information

Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records.

Publications

Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets