Identification and risk stratification of coronary disease by artificial intelligence-enabled ECG.

Background: Atherosclerotic cardiovascular disease (ASCVD) is the leading cause of death worldwide, driven primarily by coronary artery disease (CAD). ASCVD risk estimators such as the pooled cohort equations (PCE) facilitate risk stratification and primary prevention of ASCVD but their accuracy is still suboptimal. Methods: Using deep electronic health record data from 7,116,209 patients seen at 70+ hospitals and clinics across 5 states in the USA, we developed an artificial intelligence-based electrocardiogram analysis tool (ECG-AI) to detect CAD and assessed the additive value of ECG-AI-based ASCVD risk stratification to the PCE. We created independent ECG-AI models using separate neural networks including subjects without known history of ASCVD, to identify coronary artery calcium (CAC) score ≥300 Agatston units by computed tomography, obstructive CAD by angiography or procedural intervention, and regional left ventricular akinesis in ≥1 segment by echocardiogram, as a reflection of possible prior myocardial infarction (MI). These were used to assess the utility of ECG-AI-based ASCVD risk stratification in a retrospective observational study consisting of patients with PCE scores and no prior ASCVD. The study period covered all available digitized EHR data, with the first available ECG in 1987 and the last in February 2023. Findings: ECG-AI for identifying CAC ≥300, obstructive CAD, and regional akinesis achieved area under the receiver operating characteristic (AUROC) values of 0.88, 0.85, and 0.94, respectively. An ensembled ECG-AI identified 3, 5, and 10-year risk for acute coronary events and mortality independently and additively to PCE. Hazard ratios for acute coronary events over 3-years in patients without ASCVD that tested positive on 1, 2, or 3 versus 0 disease-specific ECG-AI models at cohort entry were 2.41 (2.14-2.71), 4.23 (3.74-4.78), and 11.75 (10.2-13.52), respectively. Similar stratification was observed in cohorts stratified by PCE or age. Interpretation: ECG-AI has potential to address unmet need for accessible risk stratification in patients in whom PCE under, over, or insufficiently estimates ASCVD risk, and in whom risk assessment over time periods shorter than 10 years is desired. Funding: Anumana.

Development of Electronic Health Record-Based Machine Learning Models to Predict Barrett's Esophagus and Esophageal Adenocarcinoma Risk.

INTRODUCTION: Screening for Barrett's esophagus (BE) is suggested in those with risk factors, but remains underutilized. BE/esophageal adenocarcinoma (EAC) risk prediction tools integrating multiple risk factors have been described. However, accuracy remains modest (area under the receiver-operating curve [AUROC] ≤0.7), and clinical implementation has been challenging. We aimed to develop machine learning (ML) BE/EAC risk prediction models from an electronic health record (EHR) database. METHODS: The Clinical Data Analytics Platform, a deidentified EHR database of 6 million Mayo Clinic patients, was used to predict BE and EAC risk. BE and EAC cases and controls were identified using International Classification of Diseases codes and augmented curation (natural language processing) techniques applied to clinical, endoscopy, laboratory, and pathology notes. Cases were propensity score matched to 5 independent randomly selected control groups. An ensemble transformer-based ML model architecture was used to develop predictive models. RESULTS: We identified 8,476 BE cases, 1,539 EAC cases, and 252,276 controls. The BE ML transformer model had an overall sensitivity, specificity, and AUROC of 76%, 76%, and 0.84, respectively. The EAC ML transformer model had an overall sensitivity, specificity, and AUROC of 84%, 70%, and 0.84, respectively. Predictors of BE and EAC included conventional risk factors and additional novel factors, such as coronary artery disease, serum triglycerides, and electrolytes. DISCUSSION: ML models developed on an EHR database can predict incident BE and EAC risk with improved accuracy compared with conventional risk factor-based risk scores. Such a model may enable effective implementation of a minimally invasive screening technology.

Analysis of the Effectiveness of the Ad26.COV2.S Adenoviral Vector Vaccine for Preventing COVID-19.

Importance: Continuous assessment of the effectiveness and safety of the US Food and Drug Administration-authorized SARS-CoV-2 vaccines is critical to amplify transparency, build public trust, and ultimately improve overall health outcomes. Objective: To evaluate the effectiveness of the Johnson & Johnson Ad26.COV2.S vaccine for preventing SARS-CoV-2 infection. Design, Setting, and Participants: This comparative effectiveness research study used large-scale longitudinal curation of electronic health records from the multistate Mayo Clinic Health System (Minnesota, Arizona, Florida, Wisconsin, and Iowa) to identify vaccinated and unvaccinated adults between February 27 and July 22, 2021. The unvaccinated cohort was matched on a propensity score derived from age, sex, zip code, race, ethnicity, and previous number of SARS-CoV-2 polymerase chain reaction tests. The final study cohort consisted of 8889 patients in the vaccinated group and 88 898 unvaccinated matched patients. Exposure: Single dose of the Ad26.COV2.S vaccine. Main Outcomes and Measures: The incidence rate ratio of SARS-CoV-2 infection in the vaccinated vs unvaccinated control cohorts, measured by SARS-CoV-2 polymerase chain reaction testing. Results: The study was composed of 8889 vaccinated patients (4491 men [50.5%]; mean [SD] age, 52.4 [16.9] years) and 88 898 unvaccinated patients (44 748 men [50.3%]; mean [SD] age, 51.7 [16.7] years). The incidence rate ratio of SARS-CoV-2 infection in the vaccinated vs unvaccinated control cohorts was 0.26 (95% CI, 0.20-0.34) (60 of 8889 vaccinated patients vs 2236 of 88 898 unvaccinated individuals), which corresponds to an effectiveness of 73.6% (95% CI, 65.9%-79.9%) and a 3.73-fold reduction in SARS-CoV-2 infections. Conclusions and Relevance: This study's findings are consistent with the clinical trial-reported efficacy of Ad26.COV2.S and the first retrospective analysis, suggesting that the vaccine is effective at reducing SARS-CoV-2 infection, even with the spread of variants such as Alpha or Delta that were not present in the original studies, and reaffirm the urgent need to continue mass vaccination efforts globally.

Cerebral Venous Sinus Thrombosis is not Significantly Linked to COVID-19 Vaccines or Non-COVID Vaccines in a Large Multi-State Health System.

OBJECTIVE: To assess the association of COVID-19 vaccines and non-COVID-19 vaccines with cerebral venous sinus thrombosis (CVST). MATERIALS AND METHOD: We retrospectively analyzed a cohort of 771,805 vaccination events across 266,094 patients in the Mayo Clinic Health System between 01/01/2017 and 03/15/2021. The primary outcome was a positive diagnosis of CVST, identified either by the presence of a corresponding ICD code or by an NLP algorithm which detected positive diagnosis of CVST within free-text clinical notes. For each vaccine we calculated the relative risk by dividing the incidence of CVST in the 30 days following vaccination to that in the 30 days preceding vaccination. RESULTS: We identified vaccination events for all FDA-approved COVID-19 vaccines including Pfizer-BioNTech (n = 94,818 doses), Moderna (n = 36,350 doses) and Johnson & Johnson - J&J (n = 1,745 doses). We also identified vaccinations events for 10 common FDA-approved non-COVID-19 vaccines (n = 771,805 doses). There was no statistically significant difference in the incidence rate of CVST in 30-days before and after vaccination for any vaccine in this population. We further found the baseline CVST incidence in the study population between 2017 and 2021 to be 45 to 98 per million patient years. CONCLUSIONS: This real-world evidence-based study finds that CVST is rare and is not significantly associated with COVID-19 vaccination in our patient cohort. Limitations include the rarity of CVST in our dataset, a relatively small number of J&J COVID-19 vaccination events, and the use of a population drawn from recipients of a SARS-CoV-2 PCR test in a single health system.

Mapping each pre-existing condition's association to short-term and long-term COVID-19 complications.

Understanding the relationships between pre-existing conditions and complications of COVID-19 infection is critical to identifying which patients will develop severe disease. Here, we leverage ~1.1 million clinical notes from 1803 hospitalized COVID-19 patients and deep neural network models to characterize associations between 21 pre-existing conditions and the development of 20 complications (e.g. respiratory, cardiovascular, renal, and hematologic) of COVID-19 infection throughout the course of infection (i.e. 0-30 days, 31-60 days, and 61-90 days). Pleural effusion was the most frequent complication of early COVID-19 infection (89/1803 patients, 4.9%) followed by cardiac arrhythmia (45/1803 patients, 2.5%). Notably, hypertension was the most significant risk factor associated with 10 different complications including acute respiratory distress syndrome, cardiac arrhythmia, and anemia. The onset of new complications after 30 days is rare and most commonly involves pleural effusion (31-60 days: 11 patients, 61-90 days: 9 patients). Lastly, comparing the rates of complications with a propensity-matched COVID-negative hospitalized population confirmed the importance of hypertension as a risk factor for early-onset complications. Overall, the associations between pre-COVID conditions and COVID-associated complications presented here may form the basis for the development of risk assessment scores to guide clinical care pathways.

Building a best-in-class automated de-identification tool for electronic health records through ensemble learning.

The presence of personally identifiable information (PII) in natural language portions of electronic health records (EHRs) constrains their broad reuse. Despite continuous improvements in automated detection of PII, residual identifiers require manual validation and correction. Here, we describe an automated de-identification system that employs an ensemble architecture, incorporating attention-based deep-learning models and rule-based methods, supported by heuristics for detecting PII in EHR data. Detected identifiers are then transformed into plausible, though fictional, surrogates to further obfuscate any leaked identifier. Our approach outperforms existing tools, with a recall of 0.992 and precision of 0.979 on the i2b2 2014 dataset and a recall of 0.994 and precision of 0.967 on a dataset of 10,000 notes from the Mayo Clinic. The de-identification system presented here enables the generation of de-identified patient data at the scale required for modern machine-learning applications to help accelerate medical discoveries.

Augmented curation of clinical notes from a massive EHR system reveals symptoms of impending COVID-19 diagnosis.

Understanding temporal dynamics of COVID-19 symptoms could provide fine-grained resolution to guide clinical decision-making. Here, we use deep neural networks over an institution-wide platform for the augmented curation of clinical notes from 77,167 patients subjected to COVID-19 PCR testing. By contrasting Electronic Health Record (EHR)-derived symptoms of COVID-19-positive (COVIDpos; n = 2,317) versus COVID-19-negative (COVIDneg; n = 74,850) patients for the week preceding the PCR testing date, we identify anosmia/dysgeusia (27.1-fold), fever/chills (2.6-fold), respiratory difficulty (2.2-fold), cough (2.2-fold), myalgia/arthralgia (2-fold), and diarrhea (1.4-fold) as significantly amplified in COVIDpos over COVIDneg patients. The combination of cough and fever/chills has 4.2-fold amplification in COVIDpos patients during the week prior to PCR testing, in addition to anosmia/dysgeusia, constitutes the earliest EHR-derived signature of COVID-19. This study introduces an Augmented Intelligence platform for the real-time synthesis of institutional biomedical knowledge. The platform holds tremendous potential for scaling up curation throughput, thus enabling EHR-powered early disease diagnosis.