Electronic Health Record Features Research Articles

Overtriage, Undertriage, and Value of Care after Major Surgery: An Automated, Explainable Deep Learning-Enabled Classification System.

In single-institution studies, overtriaging low-risk postoperative patients to ICUs has been associated with a low value of care; undertriaging high-risk postoperative patients to general wards has been associated with increased mortality and morbidity. This study tested the reproducibility of an automated postoperative triage classification system to generating an actionable, explainable decision support system. This longitudinal cohort study included adults undergoing inpatient surgery at two university hospitals. Triage classifications were generated by an explainable deep learning model using preoperative and intraoperative electronic health record features. Nearest neighbor algorithms identified risk-matched controls. Primary outcomes were mortality, morbidity, and value of care (inverted risk-adjusted mortality/total direct costs). Among 4,669 ICU admissions, 237 (5.1%) were overtriaged. Compared with 1,021 control ward admissions, overtriaged admissions had similar outcomes but higher costs ($15.9K [interquartile range $9.8K to $22.3K] vs $10.7K [$7.0K to $17.6K], p < 0.001) and lower value of care (0.2 [0.1 to 0.3] vs 1.5 [0.9 to 2.2], p < 0.001). Among 8,594 ward admissions, 1,029 (12.0%) were undertriaged. Compared with 2,498 control ICU admissions, undertriaged admissions had longer hospital length-of-stays (6.4 [3.4 to 12.4] vs 5.4 [2.6 to 10.4] days, p < 0.001); greater incidence of hospital mortality (1.7% vs 0.7%, p = 0.03), cardiac arrest (1.4% vs 0.5%, p = 0.04), and persistent acute kidney injury without renal recovery (5.2% vs 2.8%, p = 0.002); similar costs ($21.8K [$13.3K to $34.9K] vs $21.9K [$13.1K to $36.3K]); and lower value of care (0.8 [0.5 to 1.3] vs 1.2 [0.7 to 2.0], p < 0.001). Overtriage was associated with low value of care; undertriage was associated with both low value of care and increased mortality and morbidity. The proposed framework for generating automated postoperative triage classifications is reproducible.

An integrated LSTM-HeteroRGNN model for interpretable opioid overdose risk prediction

Opioid overdose (OD) has become a leading cause of accidental death in the United States, and overdose deaths reached a record high during the COVID-19 pandemic. Combating the opioid crisis requires targeting high-need populations by identifying individuals at risk of OD. While deep learning emerges as a powerful method for building predictive models using large scale electronic health records (EHR), it is challenged by the complex intrinsic relationships among EHR data. Further, its utility is limited by the lack of clinically meaningful explainability, which is necessary for making informed clinical or policy decisions using such models. In this paper, we present LIGHTED, an integrated deep learning model combining long short term memory (LSTM) and graph neural networks (GNN) to predict patients' OD risk. The LIGHTED model can incorporate the temporal effects of disease progression and the knowledge learned from interactions among clinical features. We evaluated the model using Cerner's Health Facts database with over 5 million patients. Our experiments demonstrated that the model outperforms traditional machine learning methods and other deep learning models. We also proposed a novel interpretability method by exploiting embeddings provided by GNNs to cluster patients and EHR features respectively, and conducted qualitative feature cluster analysis for clinical interpretations. Our study shows that LIGHTED can take advantage of longitudinal EHR data and the intrinsic graph structure of EHRs among patients to provide effective and interpretable OD risk predictions that may potentially improve clinical decision support.

A semi-supervised adaptive Markov Gaussian embedding process (SAMGEP) for prediction of phenotype event times using the electronic health record

While there exist numerous methods to identify binary phenotypes (i.e. COPD) using electronic health record (EHR) data, few exist to ascertain the timings of phenotype events (i.e. COPD onset or exacerbations). Estimating event times could enable more powerful use of EHR data for longitudinal risk modeling, including survival analysis. Here we introduce Semi-supervised Adaptive Markov Gaussian Embedding Process (SAMGEP), a semi-supervised machine learning algorithm to estimate phenotype event times using EHR data with limited observed labels, which require resource-intensive chart review to obtain. SAMGEP models latent phenotype states as a binary Markov process, and it employs an adaptive weighting strategy to map timestamped EHR features to an embedding function that it models as a state-dependent Gaussian process. SAMGEP’s feature weighting achieves meaningful feature selection, and its predictions significantly improve AUCs and F1 scores over existing approaches in diverse simulations and real-world settings. It is particularly adept at predicting cumulative risk and event counting process functions, and is robust to diverse generative model parameters. Moreover, it achieves high accuracy with few (50–100) labels, efficiently leveraging unlabeled EHR data to maximize information gain from costly-to-obtain event time labels. SAMGEP can be used to estimate accurate phenotype state functions for risk modeling research.

Prediction of new onset atrial fibrillation recurrence or persistence with artificial intelligence: first insights of the PRAFAI study

Abstract Introduction Atrial fibrillation (AF) is the most prevalent arrhythmia in the world and it is associated with a high rate of cardiovascular morbidity and mortality. A higher burden of atrial fibrillation is related to more adverse cardiovascular events and the rhythm control strategy (maintenance or recovery of sinus rhythm) is particularly indicated in patients with paroxysmal symptomatic atrial fibrillation. All things considered, in order to select the optimal therapeutic strategy, it might be useful to accurately predict the probability of recurrence or permanence of AF after its onset. The predictive models implemented so far, show a limited predictive power or were built on samples of patients undergoing specific therapies. Thus, developing an specific model based on features derived from electronic health records (EHR) of patients with a first AF episode seems essential. Purpose To develop an artificial intelligence powered predictive model based on EHR variables in order to predict the AF recurrence after its onset. Methods The methodological design of the research is based on the analysis of secondary data from a retrospective cohort obtained from the EHR of a tertiary hospital. All new onset AF from 2015 to 2018 were identified by ICD codes, and a 2 years EHR driven follow-up was made to determine the AF recurrence or permanence of the arrhythmia in those patients. A systematic review was pursued to determine which features exhibited a greater predictive power for AF recurrence, and they were inquired to the system for the mentioned participants. After the standardization of the variables and a feature selection process, several missing value imputation methods and around 12 different Machine Learning algorithms were tested to predict the AF recurrence in a subset of patients who developed AF in the first semester of 2015, in order to select the algorithms with the highest prediction accuracy for further tests. Results 523 patients had an AF onset in the first semester of 2015. Among them, 310 (59.3%) had AF recurrence in a 2 years period. A support vector machine (SVM) algorithm with a median imputation method to handle missing values performed the best and had an 0.77 AUC in the training set and 0.71 in the test set. Hyperthyroidism, vasodilator drug use and obstructive sleep apnea were notoriously related to AF recurrence, while Ic antiarrhythmic drugs, AF ablation, cognitive impairment and anxiety were the most influential variables for the absence of recurrence. Classical AF recurrence factors such as age or left atrium diameter ratify their role in this setting. Conclusion A SVM Machine Learning algorithm discloses an accurate AF recurrence probability based on EHR features. Pharmacologic and epidemiological variables, added to classically assessed ones, seem to be specially related to this arrhythmic recurrence. Further research including the whole cohort features will probably improve the accuracy of this selected SVM algorithm. Funding Acknowledgement Type of funding sources: None.

Improving the User Experience with Discount Site-Specific User Testing.

Poor electronic health record (EHR) usability is associated with patient safety concerns, user dissatisfaction, and provider burnout. EHR certification requires vendors to perform user testing. However, there are no such requirements for site-specific implementations. Health care organizations customize EHR implementations, potentially introducing usability problems. Site-specific usability evaluations may help to identify these concerns, and "discount" usability methods afford health systems a means of doing so even without dedicated usability specialists. This report characterizes a site-specific discount user testing program launched at an academic medical center. We describe lessons learned and highlight three of the EHR features in detail to demonstrate the impact of testing on implementation decisions and on users. Thirteen new EHR features which had already undergone heuristic evaluation and iterative design were evaluated over the course of three user test events. Each event included five to six users. Participants used think aloud technique. Measures of user efficiency, effectiveness, and satisfaction were collected. Usability concerns were characterized by the type of usability heuristic violated and by correctability. Usability concerns occurred at a rate of 2.5 per feature tested. Seventy percent of the usability concerns were deemed correctable prior to implementation. The first highlighted feature was moved to production despite low single ease question (SEQ) scores which may have predicted its subsequent withdrawal from production based on post implementation feedback. Another feature was rebuilt based on usability findings, and a new version was retested and moved to production. A third feature highlights an easily correctable usability concern identified in user testing. Quantitative usability metrics generally reinforced qualitative findings. Simplified user testing with a limited number of participants identifies correctable usability concerns, even after heuristic evaluation. Our discount usability approach to site-specific usability has a role in implementations and may improve the usability of the EHR for the end user.

Prediction of Postoperative Deterioration in Cardiac Surgery Patients Using Electronic Health Record and Physiologic Waveform Data.

Postoperative hemodynamic deterioration among cardiac surgical patients can indicate or lead to adverse outcomes. Whereas prediction models for such events using electronic health records or physiologic waveform data are previously described, their combined value remains incompletely defined. The authors hypothesized that models incorporating electronic health record and processed waveform signal data (electrocardiogram lead II, pulse plethysmography, arterial catheter tracing) would yield improved performance versus either modality alone. Intensive care unit data were reviewed after elective adult cardiac surgical procedures at an academic center between 2013 and 2020. Model features included electronic health record features and physiologic waveforms. Tensor decomposition was used for waveform feature reduction. Machine learning-based prediction models included a 2013 to 2017 training set and a 2017 to 2020 temporal holdout test set. The primary outcome was a postoperative deterioration event, defined as a composite of low cardiac index of less than 2.0 ml min-1 m-2, mean arterial pressure of less than 55 mmHg sustained for 120 min or longer, new or escalated inotrope/vasopressor infusion, epinephrine bolus of 1 mg or more, or intensive care unit mortality. Prediction models analyzed data 8 h before events. Among 1,555 cases, 185 (12%) experienced 276 deterioration events, most commonly including low cardiac index (7.0% of patients), new inotrope (1.9%), and sustained hypotension (1.4%). The best performing model on the 2013 to 2017 training set yielded a C-statistic of 0.803 (95% CI, 0.799 to 0.807), although performance was substantially lower in the 2017 to 2020 test set (0.709, 0.705 to 0.712). Test set performance of the combined model was greater than corresponding models limited to solely electronic health record features (0.641; 95% CI, 0.637 to 0.646) or waveform features (0.697; 95% CI, 0.693 to 0.701). Clinical deterioration prediction models combining electronic health record data and waveform data were superior to either modality alone, and performance of combined models was primarily driven by waveform data. Decreased performance of prediction models during temporal validation may be explained by data set shift, a core challenge of healthcare prediction modeling.

Exploration of machine learning methods to predict systemic lupus erythematosus hospitalizations.

Systemic lupus erythematosus (SLE) is a heterogeneous disease characterized by disease flares which can require hospitalization. Our objective was to apply machine learning methods to predict hospitalizations for SLE from electronic health record (EHR) data. We identified patients with SLE in a longitudinal EHR-based cohort with ≥2 outpatient rheumatology visits between 2012 and 2019. We applied multiple machine learning methods to predict hospitalizations with a primary diagnosis code for SLE, including decision tree, random forest, naive Bayes, logistic regression, and an ensemble method. Candidate predictors were derived from structured EHR features, including demographics, laboratory tests, medications, ICD-9/10 codes for SLE manifestations, and healthcare utilization. We used two approaches to assess these variables over longitudinal follow-up, including the incorporation of lagged features to capture changes over time of clinical data. The performance of each model was evaluated by overall accuracy, the F statistic, and the area under the receiver operator curve (AUC). We identified 1996 patients with SLE. 4.6% were hospitalized for SLE in their most recent year of follow-up. Random forest models had highest performance in predicting SLE hospitalizations, with AUC 0.751 and AUC 0.772 for two approaches (averaging and progressive), respectively. The leading predictors of SLE hospitalizations included dsDNA positivity, C3 level, blood cell counts, and inflammatory markers as well as age and albumin. We have demonstrated that machine learning methods can predict SLE hospitalizations. We identified key predictors of these events including known markers of SLE disease activity; further validation in external cohorts is warranted.

Revisiting Provider Communication to Support Team Cohesiveness: Implications for Practice, Provider Burnout, and Technology Application in Primary Care Settings.

Background Effective team communication is an essential aspect of care delivery and the coordination of patients in primary care settings. With the rapid evolution of health information technology (HIT), including the implementation of electronic health records, there remains a gap in the literature about preferred methods of primary care team communication and the subsequent impact of provider and team outcomes (e.g., team cohesiveness; burnout). This study explores the impact of varying modes of communication across provider disciplines and by geographic settings during primary care delivery. Methods We used a cross-sectional survey design to collect data from a random convenience sample of PCPs (physicians, nurse practitioners, and physician assistants) (n = 314) in New York State (NYS). We mailed a paper survey with validated measures for communication methods, team cohesiveness, and provider outcomes (burnout, job dissatisfaction, and the intention to leave position). Descriptive statistics, linear regression models, and crude and adjusted odds ratios while controlling for individual and practice characteristics were calculated. Results In-person communication was found to yield greater job satisfaction and less intention to leave current position in the next year (p=0.02) compared to other forms of communication including electronic health record features. The odds of job satisfaction was 1.51 times higher with in-person communication (OR: 1.51, 95% CI: 1.05, 2.19), and the odds of intending to leave a position was 45% less with in-person communication (OR: 0.55, 95% CI: 0.36, 0.85). The odds of reporting burnout at work was 36% less with in-person communication (OR: 0.64, 95% CI: 0.43, 0.92) compared to other communication modalities. There was no significant association between team communication via the EHR and team cohesiveness, provider burnout, or job satisfaction. Conclusion This study demonstrates evidence that in-person communication is more likely to reduce burnout and job dissatisfaction compared to other forms of communication infrastructure in primary care settings. More research is needed to understand PCP perspectives about the functionality and potential burden that inhibits the use of EHR features for provider-provider communication. In addition, attention to the needs of teams by geographic location and by workforce discipline is warranted to ensure effective HIT communication application adoption.

Opportunities to use electronic health record audit logs to improve cancer care.

The rapid adoption of electronic health records (EHRs) has created extensive repositories of digitized data that can be used to inform improvements in care delivery, processes, and patient outcomes. While the clinical data captured in EHRs are widely used for such efforts, EHRs also capture audit log data that reflect how users interact with the EHR to deliver care. Automatically collected audit log data provide a unique opportunity for new insights into EHR user behavior and decision‐making processes. Here, we provide an overview of audit log data and examples that could be used to improve oncology care and outcomes in four domains: diagnostic reasoning and consumption, care team collaboration and communication, patient outcomes and experience, and provider burnout/fatigue. This data source could identify gaps in performance and care, physician uptake of EHR features that enhance decision‐making, and integration of data trends for oncology. Ensuring researchers and oncologists are familiar with the data's potential and developing the data engineering capacity to utilize this rich data source, will expand the breadth of research to improve cancer care.

Development and Structure of an Accurate Machine Learning Algorithm to Predict Inpatient Mortality and Hospice Outcomes in the Coronavirus Disease 2019 Era.

Background:The coronavirus disease 2019 (COVID-19) pandemic has challenged the accuracy and racial biases present in traditional mortality scores. An accurate prognostic model that can be applied to hospitalized patients irrespective of race or COVID-19 status may benefit patient care.Research Design:This cohort study utilized historical and ongoing electronic health record features to develop and validate a deep-learning model applied on the second day of admission predicting a composite outcome of in-hospital mortality, discharge to hospice, or death within 30 days of admission. Model features included patient demographics, diagnoses, procedures, inpatient medications, laboratory values, vital signs, and substance use history. Conventional performance metrics were assessed, and subgroup analysis was performed based on race, COVID-19 status, and intensive care unit admission.Subjects:A total of 35,521 patients hospitalized between April 2020 and October 2020 at a single health care system including a tertiary academic referral center and 9 community hospitals.Results:Of 35,521 patients, including 9831 non-White patients and 2020 COVID-19 patients, 2838 (8.0%) met the composite outcome. Patients who experienced the composite outcome were older (73 vs. 61 y old) with similar sex and race distributions between groups. The model achieved an area under the receiver operating characteristic curve of 0.89 (95% confidence interval: 0.88, 0.91) and an average positive predictive value of 0.46 (0.40, 0.52). Model performance did not differ significantly in White (0.89) and non-White (0.90) subgroups or when grouping by COVID-19 status and intensive care unit admission.Conclusion:A deep-learning model using large-volume, structured electronic health record data can effectively predict short-term mortality or hospice outcomes on the second day of admission in the general inpatient population without significant racial bias.

A Free, Open-Source, Offline Digital Health System for Refugee Care.

BackgroundRise of conflict, extreme weather events, and pandemics have led to larger displaced populations worldwide. Displaced populations have unique acute and chronic health needs that must be met by low-resource health systems. Electronic health records (EHRs) have been shown to improve health outcomes in displaced populations, but need to be adapted to meet the constraints of these health systems.ObjectiveThe aim of this viewpoint is to describe the development and deployment of an EHR designed to care for displaced populations in low-resource settings.MethodsUsing a human-centered design approach, we conducted in-depth interviews and focus groups with patients, health care providers, and administrators in Lebanon and Jordan to identify the essential EHR features. These features, including modular workflows, multilingual interfaces, and offline-first capabilities, led to the development of the Hikma Health EHR, which has been deployed in Lebanon and Nicaragua.ResultsWe report the successes and challenges from 12 months of Hikma Health EHR deployment in a mobile clinic providing care to Syrian refugees in Bekaa Valley, Lebanon. Successes include the EHR’s ability to (1) increase clinical efficacy by providing detailed patient records, (2) be adaptable to the threats of COVID-19, and (3) improve organizational planning. Lessons learned include technical fixes to methods of identifying patients through name or their medical record ID.ConclusionsAs the number of displaced people continues to rise globally, it is imperative that solutions are created to help maximize the health care they receive. Free, open-sourced, and adaptable EHRs can enable organizations to better provide for displaced populations.

Computational signatures for post-cardiac arrest trajectory prediction: Importance of early physiological time series

BackgroundThere is an unmet need for timely and reliable prediction of post-cardiac arrest (CA) clinical trajectories. We hypothesized that physiological time series (PTS) data recorded on the first day of intensive care would contribute significantly to discrimination of outcomes at discharge. Patients and methodsAdult patients in the multicenter eICU database who were mechanically ventilated after resuscitation from out-of-hospital CA were included. Outcomes of interest were survival, neurological status based on Glasgow motor subscore (mGCS) and surrogate functional status based on discharge location (DL), at hospital discharge. Three machine learning predictive models were trained, one with features from the electronic health records (EHR), the second using features derived from PTS collected in the first 24 h after ICU admission (PTS24), and the third combining PTS24 and EHR. Model performances were compared, and the best performing model was externally validated in the MIMIC-III dataset. ResultsData from 2216 admissions were included in the analysis. Discrimination of prediction models combining EHR and PTS24 features was higher than models using either EHR or PTS24 for prediction of survival (AUROC 0.83, 0.82 and 0.79 respectively), neurological outcome (0.87, 0.86 and 0.79 respectively), and DL (0.80, 0.78 and 0.76 respectively). External validation in MIMIC-III (n = 86) produced similar model performance. Feature analysis suggested prognostic significance of previously unknown EHR and PTS24 variables. ConclusionThese results indicate that physiological data recorded in the early phase after CA resuscitation contain signatures that are linked to post-CA outcome. Additionally, they attest to the effectiveness of ML for post-CA predictive modeling.

Detecting early signs of Alzheimer’s disease and related dementia onset from the EHR

As dementia is widely under-detected, a predictive model using electronic health records (EHR) could provide a method for early screening to implement preventive strategies. There is limited research on using EHR to identify persons with Alzheimer’s disease (AD) and related dementias (RD). In a data-driven approach, we used all ICD-9 diagnosis and CPT procedure codes from statewide inpatient, ambulatory surgery, and Medicare records, in addition to age at baseline and gender, to detect AD/RD from the Cache County Study on Memory in Aging (1995–2009). After removing participants diagnosed with dementia at baseline (n=335), 3882 (82%) Cache County Study participants could be linked to inpatient, ambulatory surgery, and/or Medicare EHR records; 484 (12.5%) of these 3882 had incident all-cause dementia, with 308 (7.9%) having AD/AD comorbid with RD; and 176 (4.5%) having RD without AD. We removed participant’s ICD-9 codes occurring after first AD/RD diagnoses. EHR features (~2000) along with gold-standard diagnoses as class labels were then used to train and detect AD and/or RD using a Gradient Boosting Trees machine learning algorithm. Models evaluated with nested cross-validation yielded AUCs of 0.70 for all-cause dementia, 0.69 for AD/AD comorbid with RD, and 0.67 for RD without AD. Key factors detecting AD/RD included age at enrollment, cardiovascular, metabolic, and kidney disease, and sleep disturbances, with feature importance varying by record type and time frame prior to dementia onset. Our findings suggest that a patient’s health status up to 12 years prior may be useful in identifying individuals at-risk for dementia development.

Association of Postoperative Undertriage to Hospital Wards With Mortality and Morbidity

Undertriaging patients who are at increased risk for postoperative complications after surgical procedures to low-acuity hospital wards (ie, floors) rather than highly vigilant intensive care units (ICUs) may be associated with risk of unrecognized decompensation and worse patient outcomes, but evidence for these associations is lacking. To test the hypothesis that postoperative undertriage is associated with increased mortality and morbidity compared with risk-matched ICU admission. This longitudinal cross-sectional study was conducted using data from the University of Florida Integrated Data Repository on admissions to a university hospital. Included patients were individuals aged 18 years or older who were admitted after a surgical procedure from June 1, 2014, to August 20, 2020. Data were analyzed from April through August 2021. Ward admissions were considered undertriaged if their estimated risk for hospital mortality or prolonged ICU stay (ie, ≥48 hours) was in the top quartile among all inpatient surgical procedures according to a validated machine-learning model using preoperative and intraoperative electronic health record features available at surgical procedure end time. A nearest neighbors algorithm was used to identify a risk-matched control group of ICU admissions. The primary outcomes of hospital mortality and morbidity were compared among appropriately triaged ward admissions, undertriaged wards admissions, and a risk-matched control group of ICU admissions. Among 12 348 postoperative ward admissions, 11 042 admissions (89.4%) were appropriately triaged (5927 [53.7%] women; median [IQR] age, 59 [44-70] years) and 1306 admissions (10.6%) were undertriaged and matched with a control group of 2452 ICU admissions. The undertriaged group, compared with the control group, had increased median [IQR] age (64 [54-74] years vs 62 [50-73] years; P = .001) and increased proportions of women (649 [49.7%] women vs 1080 [44.0%] women; P < .001) and admitted patients with do not resuscitate orders before first surgical procedure (53 admissions [4.1%] vs 27 admissions [1.1%]); P < .001); 207 admissions that were undertriaged (15.8%) had subsequent ICU admission. In the validation cohort, hospital mortality and prolonged ICU stay estimations had areas under the receiver operating characteristic curve of 0.92 (95% CI, 0.91-0.93) and 0.92 (95% CI, 0.92-0.92), respectively. The undertriaged group, compared with the control group, had similar incidence of prolonged mechanical ventilation (32 admissions [2.5%] vs 53 admissions [2.2%]; P = .60), decreased median (IQR) total costs for admission ($26 900 [$18 400-$42 300] vs $32 700 [$22 700-$48 500]; P < .001), increased median (IQR) hospital length of stay (8.1 [5.1-13.6] days vs 6.0 [3.3-9.3] days, P < .001), and increased incidence of hospital mortality (19 admissions [1.5%] vs 17 admissions [0.7%]; P = .04), discharge to hospice (23 admissions [1.8%] vs 14 admissions [0.6%]; P < .001), unplanned intubation (45 admissions [3.4%] vs 49 admissions [2.0%]; P = .01), and acute kidney injury (341 admissions [26.1%] vs 477 admissions [19.5%]; P < .001). This study found that admitted patients at increased risk for postoperative complications who were undertriaged to hospital wards had increased mortality and morbidity compared with a risk-matched control group of admissions to ICUs. Postoperative undertriage was identifiable using automated preoperative and intraoperative data as features in real-time machine-learning models.

Patient voice and health education for familial hypercholesterolaemia

Objective: Familial hypercholesterolaemia (FH), an autosomal dominant disorder causing elevated low-density lipoprotein (LDL) cholesterol from birth resulting in premature cardiovascular disease, is only diagnosed in 10% of affected patients. This study involved partnering with patients with FH and with primary care providers (PCPs) to understand health priorities and translate them into hypotheses for future research and enhancement of health practices via electronic health records (EHRs). The goal was to strengthen genetic health education for clinicians and for patients and their families, including improved diagnosis, knowledge and treatment. Perceptions regarding genetic health education and healthcare related to FH facilitated by the use of an EHR for diagnosis and treatment have not been studied. Design: Mixed-methods exploratory qualitative research and surveys. Setting: Qualitative research included five focus groups, 34 semi-structured key informant interviews and open-ended survey items with patients and PCPs at a large medical centre in Western New York. Method: Data were thematically coded to identify themes as formative work for the improvement of relevant EHR features, diagnosis, treatment and genetic health education via information sharing between clinicians and patients. Results: Themes included genetic health knowledge; the importance of being diagnosed; communication between patients, family members and medical professionals; outreach via patients’ own advocacy; and treatment, technology, motivation, trust, outside resources (for further genetic health education and support) and awareness of effective treatments. Conclusion: Patients and clinicians can contribute to the development of EHR support for the genetic health education of patients and their families, and for improved diagnosis and treatment of FH. Using their ideas in the development of effective strategies could improve the currently low rate of FH diagnosis and cascade screening (for family members), as well as enhance physician and patient genetic health knowledge and self-empowerment.

Electronic health records features associated to coronary artery disease in patients with chest pain

Abstract Introduction Cardiovascular diseases remain the main cause of death in most European countries and coronary artery disease (CAD) follows an ascendent tendency. Current guidelines on chronic coronary syndromes recommend to estimate the CAD pre-test probability based on age, sex and symptoms in patients with suspected CAD. Besides, they underline the importance of adapting this likelihood with the patient information, although the risk the physician attributes to the medical history might not be concordant among colleagues. Electronic health records (EHR) are the main patient-level data collection tools in clinical care and data capture systems compile this clinical information into research-oriented datasets. We can analyze which variables in EHR have stronger associations with CAD in this context to facilitate a more concise and personalized adjustment of the pre-test CAD risk. Purpose To analyze the association between features in EHR with the development of CAD in patients who attended the local public health system for thoracic pain in the 2016–2018 period. Methods We queried the system for patients who required medical attendance for thoracic pain or equivalent symptoms from 2016 to 2018 in the public health system. We extracted data within this subset of the potentially predictive features according to a literature review. We investigated whether patients developed CAD in the year following their attendance. A database cleaning process removed subjects without further diagnostic tests. We used parametric tests to assess the distribution differences of these features between patients who developed CAD and those who did not and we applied logistic regression to document their prediction potential. Results The EHR dataset extracted from the system contained information about 2,199,711 individuals, out of which 43,835 required medical attention for thoracic pain in the selected period. 10,463 patients had no further diagnostic tests, so were excluded. Among the remaining 33,372 patients, 5,379 (16.1%) developed CAD in the follow-up period. Heart failure and traditional cardiovascular risk factors were predominant in the CAD group, and so were related blood test features and therapies. Any history of CAD diagnostic test was significantly associated to CAD presentation. Ischemic etiology of the chest pain was more unlikely in patients with anxiety history and benzodiazepine prescription. Proton pump inhibitors use and hearth rhythm disorders did not increase CAD risk in adjusted models. Conclusions Medical history, basal blood tests and treatment information available in EHR are related to CAD risk in patients with chest pain. Considering this information seems valuable for a personalized risk adjustment of the CAD pre-test probability in symptomatic patients. Further studies with multivariable models and machine learning techniques applied to this dataset might entail more specific predictive algorithms. Funding Acknowledgement Type of funding sources: None.

Influencing factors of Barthel index scores among the community-dwelling elderly in Hong Kong: a random intercept model

BackgroundBarthel Index (BI) is one of the most widely utilized tools for assessing functional independence in activities of daily living. Most existing BI studies used populations with specific diseases (e.g., Alzheimer’s and stroke) to test prognostic factors of BI scores; however, the generalization of these findings was limited when the target populations varied.ObjectivesThe aim of the present study was to utilize electronic health records (EHRs) and data mining techniques to develop a generic procedure for identifying prognostic factors that influence BI score changes among community-dwelling elderly.MethodsLongitudinal data were collected from 113 older adults (81 females; mean age = 84 years, SD = 6.9 years) in Hong Kong elderly care centers. Visualization technologies were used to align annual BI scores with individual EHRs chronologically. Linear mixed-effects (LME) regression was conducted to model longitudinal BI scores based on socio-demographics, disease conditions, and features extracted from EHRs.ResultsThe visualization presented a decline in BI scores changed by time and health history events. The LME model yielded a conditional R2 of 84%, a marginal R2 of 75%, and a Cohen’s f2 of 0.68 in the design of random intercepts for individual heterogeneity. Changes in BI scores were significantly influenced by a set of socio-demographics (i.e., sex, education, living arrangement, and hobbies), disease conditions (i.e., dementia and diabetes mellitus), and EHRs features (i.e., event counts in allergies, diagnoses, accidents, wounds, hospital admissions, injections, etc.).ConclusionsThe proposed visualization approach and the LME model estimation can help to trace older adults’ BI score changes and identify the influencing factors. The constructed long-term surveillance system provides reference data in clinical practice and help healthcare providers manage the time, cost, data and human resources in community-dwelling settings.

Underuse of Electronic Health Record Features—The Case for CancelRx

Underuse of Electronic Health Record Features—The Case for CancelRx

Improving the Efficiency of Clinical Trial Recruitment Using an Ensemble Machine Learning to Assist With Eligibility Screening.

ObjectiveEfficiently identifying eligible patients is a crucial first step for a successful clinical trial. The objective of this study was to test whether an approach using electronic health record (EHR) data and an ensemble machine learning algorithm incorporating billing codes and data from clinical notes processed by natural language processing (NLP) can improve the efficiency of eligibility screening.MethodsWe studied patients screened for a clinical trial of rheumatoid arthritis (RA) with one or more International Classification of Diseases (ICD) code for RA and age greater than 35 years, from a tertiary care center and a community hospital. The following three groups of EHR features were considered for the algorithm: 1) structured features, 2) the counts of NLP concepts from notes, 3) health care utilization. All features were linked to dates. We applied random forest and logistic regression with least absolute shrinkage and selection operator penalty against the following two standard approaches: 1) one or more RA ICD code and no ICD codes related to exclusion criteria (ScreenRAICD1 +EX) and 2) two or more RA ICD codes (ScreenRAICD2). To test the portability, we trained the algorithm at one institution and tested it at the other.ResultsIn total, 3359 patients at Brigham and Women’s Hospital (BWH) and 642 patients at Faulkner Hospital (FH) were studied, with 461 (13.7%) eligible patients at BWH and 84 (13.4%) at FH. The application of the algorithm reduced ineligible patients from chart review by 40.5% at the tertiary care center and by 57.0% at the community hospital. In contrast, ScreenRAICD2 reduced patients for chart review by 2.7% to 11.3%; ScreenRAICD1+EX reduced patients for chart review by 63% to 65% but excluded 22% to 27% of eligible patients.ConclusionThe ensemble machine learning algorithm incorporating billing codes and NLP data increased the efficiency of eligibility screening by reducing the number of patients requiring chart review while not excluding eligible patients. Moreover, this approach can be trained at one institution and applied at another for multicenter clinical trials.

Web Services and Cloud Computing in Pediatric Care.

Electronic health record (EHR) systems do not uniformly implement pediatric-supportive functionalities. One method of adding these capabilities across EHR platforms is to integrate Web services and Web applications that may perform decision support and store data in the cloud when the EHR platform is able to integrate Web services. Specific examples of these services are described, such as immunization clinical decision support services, consumer health resources, and bilirubin nomograms. Health care providers, EHR vendors, and developers share responsibilities in the appropriate development, integration, and use of Web services and Web applications as they relate to best practices in the areas of data security and confidentiality, technical availability, audit trails, terminology and messaging standards, compliance with the Health Insurance Portability and Accountability Act, testing, usability, and other considerations. It is desirable for health care providers to have knowledge of Web services and Web applications that can improve pediatric capabilities in their own EHRs because this will naturally inform discussions concerning EHR features and facilitate implementation and subsequent use of these capabilities by clinicians caring for children.