While static risk models may identify key driving risk factors, the dynamic nature of risk requires up-to-date risk information to guide treatment decision making. Bleeding is a complication of percutaneous coronary intervention (PCI), and existing risk models produce only a single risk estimate anchored at a single point in time, despite the dynamic nature of this risk. Using data available from the National Cardiovascular Data Registry (NCDR) CathPCI, we trained 6 different tree-based machine learning models to estimate the risk of bleeding at key decision points: 1) choice of access site, 2) prescription of medication before PCI, and 3) choice of closure device. We began with 3,423,170 PCIs performed between July 2009 through April 2015. We included only index PCIs and removed anyone who had missing data regarding bleeding events or underwent coronary artery bypass grafting during the index admission. We included 2,868,808 PCIs; 2,314,446 (80.7%) before 2014 for training and 554,362 (19.3%) remaining for validation. This study considered all data available from the Registry prior to patient discharge: patient characteristics, coronary anatomy and lesion characterization, laboratory data, past medical history, anti-coagulation, stent type, and closure method categories. The primary outcome was any in-hospital bleeding event within 72 hours after the start of the PCI procedure. Discrimination improved from an area under the receiver operating characteristic curve (AUROC) of 0.812 using only presentation variables to 0.845 using all variables. Among 123,712 patients classified as low risk by the initial model, 14,441 were reclassified as moderate risk (1.4% experienced bleeds), while 723 were reclassified as high risk (12.5% experienced bleeds). Static risk prediction models have more predictive error than those that update risk prediction with newly available data, which provides up-to-date risk prediction for individualized care throughout a hospitalization.

Background: Atrial fibrillation (AF) is a leading cause of cardiovascular morbidity and mortality, and early detection is important for stroke prevention. The Apple Heart Study (AHS) demonstrated that wearable devices capable of irregular pulse notifications (IPN) can aid in identifying AF. The rates of subsequent diagnoses of cardiovascular disease (CVD) remain unknown. Hypothesis: We aimed to evaluate how smartwatch-detected IPNs are associated with key cardiovascular diagnoses. We hypothesized that, among AHS participants, receiving IPNs would be associated with increased odds of subsequently reporting major cardiovascular diagnoses compared to participants without IPNs. Methods: Out of 419,297 enrolled AHS participants, 288,533 (69%) individuals who completed the end-of-study survey (EOS) were included in this analysis. The self-reported diagnoses consisted of heart failure (HF), stroke, transient ischemic attack (TIA), myocardial infarction (MI), and pulmonary embolism (PE), considered individually and as a composite (MACE). Logistic regression was used to estimate odds ratios (ORs) for each diagnosis among IPN versus non-IPN participants, adjusted for sex, age, and CHA2DS2-VASc components. Results: The EOS survey was completed by 908 notified participants (42% among the IPN cohort) and 287,625 non-notified participants (69% among the non-IPN cohort). IPN recipients were older, more often male, and had a higher comorbidity burden and CVD risk factors than the total cohort (Table 1). The IPN cohort reported higher rates of HF, MI, stroke, TIA, and PE compared to those who did not receive an IPN (Table 2). The fully adjusted logistic regression analysis showed increased odds of each diagnosis among the IPN group (Figure 1). Conclusions: Participants with an IPN had higher odds of subsequently reporting major adverse cardiovascular diagnoses. These associations may reflect pre-existing conditions, underlying CVD risk, and comorbidities among participants who received an IPN, which has a high positive predictive value (PPV) for detecting AF. Given the elevated odds and high PPV, IPNs may warrant clinical consideration of evaluation for CVD or comorbidities, and future studies should assess the diagnostic and management pathways following IPNs.

Objective To evaluate differences in study engagement in diverse racial/ethnic groups that have been significantly underrepresented in atrial fibrillation and digital clinical trials. Patients and Methods This was a secondary analysis of participants from the Apple Heart Study, a prospective, siteless, single-arm pragmatic clinical trial from November 29, 2017, to January 31, 2019. Black, Hispanic, Asian, and White participants were monitored using an irregular rhythm notification algorithm designed to detect atrial fibrillation on a smartwatch. Logistic regression was performed to evaluate the relationship between race/ethnicity and completion of the first study visit after an irregular rhythm notification, adjusting for demographic characteristics and comorbidities. Results Of the 419,297 participants, 393,396 (93.8%) individuals self-identified as White, Black, Hispanic, or Asian. Overall, participants were 57% men and had a mean (SD) age of 41 (13) years. Among 2044 (0.52%) participants who received an irregular rhythm notification, non-White participants had lower odds of completing the initial virtual study visit compared with White participants (Black: OR, 0.61; 95% CI, 0.39-0.94; Hispanic: OR, 0.62; 95% CI, 0.40-0.95; Asian: OR, 0.40; 95% CI, 0.23-0.66) after multivariate adjustment. Among those who completed the initial study visit, there was no statistically significant difference in the odds of returning the electrocardiogram patch in the non-White groups compared with that of the White group. Conclusion Despite successful recruitment of racially and ethnically diverse participants, there were differences in subsequent engagement by non-White compared with that by White participants. Equitable representation and engagement of diverse racial and ethnic groups in digital clinical studies requires further study. Trial Registration Clinicaltrials.gov Identifier: NCT03335800

This article reviews the use of digital wearable technologies for monitoring of three common cardiovascular conditions: hypertension, heart failure, and atrial fibrillation.

Implantable cardioverter defibrillators (ICDs) are strongly recommended for primary prevention in patients with symptomatic heart failure and an ejection fraction less than 35% if meaningful survival greater than one year is expected. However, currently many people die within one year of ICD implantation. Current prediction models for death following primary prevention ICD placement have only modest discrimination.

The digital clinical trial is fast emerging as a pragmatic trial that can improve a trial's design including recruitment and retention, data collection and analytics. To that end, digital platforms such as electronic health records or wearable technologies that enable passive data collection can be leveraged, alleviating burden from the participant and study coordinator. However, there are challenges. For example, many of these data sources not originally intended for research may be noisier than traditionally obtained measures. Further, the secure flow of passively collected data and their integration for analysis is non-trivial. The Apple Heart Study was a prospective, single-arm, site-less digital trial designed to evaluate the ability of an app to detect atrial fibrillation. The study was designed with pragmatic features, such as an app for enrollment, a wearable device (the Apple Watch) for data collection, and electronic surveys for participant-reported outcomes that enabled a high volume of patient enrollment and accompanying data. These elements led to challenges including identifying the number of unique participants, maintaining participant-level linkage of multiple complex data streams, and participant adherence and engagement. Novel solutions were derived that inform future designs with an emphasis on data management. We build upon the excellent framework of the Clinical Trials Transformation Initiative to provide a comprehensive set of guidelines for data management of the digital clinical trial that include an increased role of collaborative data scientists in the design and conduct of the modern digital trial.

Background:Current risk scores that are solely based on clinical factors have shown modest predictive ability for understanding of factors associated with gaps in real-world prescription of oral anticoagulation (OAC) in patients with atrial fibrillation (AF). Objective:In this study, we sought to identify the role of social and geographic determinants, beyond clinical factors associated with variation in OAC prescriptions using a large national registry of ambulatory patients with AF. Methods:Between January 2017 and June 2018, we identified patients with AF from the American College of Cardiology PINNACLE (Practice Innovation and Clinical Excellence) Registry. We examined associations between patient and site-of-care factors and prescription of OAC across U.S. counties. Several machine learning (ML) methods were used to identify factors associated with OAC prescription. Results:Among 864,339 patients with AF, 586,560 (68%) were prescribed OAC. County OAC prescription rates ranged from 26.8% to 93%, with higher OAC use in the Western United States. Supervised ML analysis in predicting likelihood of OAC prescriptions and identified a rank order of patient features associated with OAC prescription. In the ML models, in addition to clinical factors, medication use (aspirin, antihypertensives, antiarrhythmic agents, lipid modifying agents), and age, household income, clinic size, and U.S. region were among the most important predictors of an OAC prescription. Conclusion:In a contemporary, national cohort of patients with AF underuse of OAC remains high, with notable geographic variation. Our results demonstrated the role of several important demographic and socioeconomic factors in underutilization of OAC in patients with AF.

An app-based clinical trial enrolment process can contribute to duplicated records, carrying data management implications. Our objective was to identify duplicated records in real time in the Apple Heart Study (AHS). We leveraged personal identifiable information (PII) to develop a dissimilarity score (DS) using the Damerau-Levenshtein distance. For computational efficiency, we focused on four types of records at the highest risk of duplication. We used the receiver operating curve (ROC) and resampling methods to derive and validate a decision rule to classify duplicated records. We identified 16,398 (4%) duplicated participants, resulting in 419,297 unique participants out of a total of 438,435 possible. Our decision rule yielded a high positive predictive value (96%) with negligible impact on the trial's original findings. Our findings provide principled solutions for future digital trials. When establishing deduplication procedures for digital trials, we recommend collecting device identifiers in addition to participant identifiers; collecting and ensuring secure access to PII; conducting a pilot study to identify reasons for duplicated records; establishing an initial deduplication algorithm that can be refined; creating a data quality plan that informs refinement; and embedding the initial deduplication algorithm in the enrolment platform to ensure unique enrolment and linkage to previous records.

BACKGROUND:Optimal antithrombotic management of patients with preexisting atrial fibrillation undergoing transcatheter aortic valve replacement is challenging given the need to balance the risk of bleeding and thromboembolism. We aimed to examine variation in care and association of antithrombotic therapies with 1-year outcomes of stroke, bleeding, and mortality in patients undergoing transcatheter aortic valve replacement with concomitant atrial fibrillation in the United States. METHODS:Patients who underwent transcatheter aortic valve replacement with preexisting atrial fibrillation from November 2011 through September 2015 in the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy registry linked to the Medicare database were examined according to receipt of oral anticoagulants (OACs) or antiplatelet therapies (APTs) or a combination of these (OAC+APT) at discharge. To assess the associations of antithrombotic therapies with 1-year outcomes of stroke, bleeding, and mortality, we utilized inverse probability weighting for antithrombotic therapies and multivariable regression modeling to adjust for patient- and hospital-level variables. RESULTS:In the 11 382 patients included in our study, 5833 (51.2%) were discharged on OAC+APT, 4786 (42.0%) on APT alone, and 763 (6.7%) on OAC alone. There was significant variability in discharge medication patterns, including 42% of patients discharged without OAC therapy. In adjusted analyses, the risk for all-cause mortality and stroke was not significantly different when comparing the 3 different antithrombotic strategies. Risk of bleeding was higher with OAC+APT compared with APT alone (hazard ratio, 1.16 [95% CI, 1.05–1.27]) and similar compared with OAC alone (hazard ratio, 1.17 [95% CI, 0.93–1.47]). CONCLUSIONS:There was significant variability in discharge medication patterns across US sites in patients with atrial fibrillation undergoing transcatheter aortic valve replacement, including significant underuse of OAC in this high-risk cohort. The use of OAC+APT (versus OAC alone or APT alone) was not associated with a lower risk of stroke or mortality but was associated with increased risk of bleeding complications at 1 year compared with APT alone.

Before the coronavirus disease 2019 (COVID-19) pandemic, use of telehealth services had been limited in cardiovascular care. Potential benefits of telehealth include improved access to care, more efficient care management, reduced costs, the ability to assess patients within their homes while involving key caretakers in medical decisions, maintaining social distance, and increased patient satisfaction. Challenges include changes in payment models, issues with data security and privacy, potential depersonalization of the patient-clinician relationship, limitations in the use of digital health technologies, and the potential impact on disparities, including socioeconomic, gender, and age-related issues and access to technology and broadband. Implementation and expansion of telehealth from a policy and reimbursement practice standpoint are filled with difficult decisions, yet addressing these are critical to the future of health care.

BACKGROUND:New methods such as machine learning techniques have been increasingly used to enhance the performance of risk predictions for clinical decision-making. However, commonly reported performance metrics may not be sufficient to capture the advantages of these newly proposed models for their adoption by health care professionals to improve care. Machine learning models often improve risk estimation for certain subpopulations that may be missed by these metrics.METHODS AND RESULTS:This article addresses the limitations of commonly reported metrics for performance comparison and proposes additional metrics. Our discussions cover metrics related to overall performance, discrimination, calibration, resolution, reclassification, and model implementation. Models for predicting acute kidney injury after percutaneous coronary intervention are used to illustrate the use of these metrics.CONCLUSIONS:We demonstrate that commonly reported metrics may not have sufficient sensitivity to identify improvement of machine learning models and propose the use of a comprehensive list of performance metrics for reporting and comparing clinical risk prediction models.

Background: The risk of acute kidney injury (AKI) for percutaneous coronary intervention (PCI) can inform decision-making and mitigation. However, risk assessment is hampered by the burden of manually abstracting and inputting data required by the established model developed on registry data. Accordingly, there is a need to develop AKI models that use automatically extracted electronic health records (EHR) data and evaluate their performance in comparison to the contemporary registry model. Methods: We used EHR data for PCIs performed at Yale New Haven Hospital (YNHH) and the linked local National Cardiovascular Data Registry (NCDR) CathPCI data to develop and compare the EHR models with the NCDR model. AKI was defined as an increase of ≥0.3 mg/dL or 50% in serum creatinine. To maximize interoperability across systems and usability at the point of care, we only considered structured EHR readily extractable before PCI, and implemented simple but universal data preprocessing. Longitudinal data was aggregated into 24h, 7d and 14d prior to PCI. We used Lasso regression and gradient descent boosting (GDB) to train the models and tested their capacity to approximate the NCDR model predictions and predict AKI independently. We evaluated performance via cross-validation by c-statistic, calibration slope, and predictive range defined by event rate difference in the lowest and highest risk deciles. We used paired t tests for model comparison. Results: The cohort included 9,202 PCIs from December 2012 to November 2019 at YNHH for 8,445 patients and had a mean age of 67.0±12.0 years with 27.6% females and 798 (8.7%) AKIs. Evaluated on the same test sets, the EHR models approximating the NCDR model predictions achieved similar performance to that model (all p>0.05). Further, the EHR model employing GDB to independently predict AKI achieved slightly better performance than the NCDR model in c-statistic (0.84 vs 0.83, p=0.02) and predictive range (41.2% vs 39.5%, p=0.13) with similar calibration (slope 0.98 vs 0.99). Conclusion: The automatically extracted EHR data produced models with similar performance to the registry model. These models can automate AKI risk assessment for PCI, alleviating provider burden and increase uptake of risk assessment.

BACKGROUND:Bleeding avoidance strategies (BASs) are increasingly adopted for patients undergoing percutaneous coronary intervention (PCI) due to bleeding complications. However, their association with bleeding events outside of Western countries remains unclear. In collaboration with the National Cardiovascular Data Registry (NCDR) CathPCI registry, we aimed to assess the time trend and impact of BAS utilization among Japanese patients.METHODS:Our study included 19,656 consecutive PCI patients registered over 10 years. These patients were divided into 4-time frame groups (T1: 2008-2011, T2: 2012-2013, T3: 2014-2015, and T4: 2016-2018). BAS was defined as the use of transradial approach or vascular closure device (VCD) use after transfemoral approach (TFA). Model performance of the NCDR CathPCI bleeding model was evaluated. The degree of bleeding reduction associated with BAS adoption was estimated via multilevel mixed-effects multivariable logistic regression analysis.RESULTS:The NCDR CathPCI bleeding risk score demonstrated good discrimination in the Japanese population (C-statistics 0.79-0.81). The BAS adoption rate increased from 43% (T1) to 91% (T4), whereas the crude CathPCI-defined bleeding rate decreased from 10% (T1) to 7% (T4). Adjusted odds ratios for bleeding events were 0.25 (95% confidence interval, 0.14-0.45, P< .001) for those undergoing TFA with VCD in T4 and 0.26 (95% confidence interval 0.20-0.35, P< .001) for transradial approach in T4 compared to patients that received TFA without VCD in T1.CONCLUSIONS:BAS use over the studied time frames was associated with lower risk of bleeding complications among Japanese. Nonetheless, observed bleeding rates remained higher compared to the US population.

This cohort study investigates the association of wearable device use with pulse rate and health care use among adults with atrial fibrillation (AF).

Importance Accurate prediction of adverse outcomes after acute myocardial infarction (AMI) can guide the triage of care services and shared decision-making, and novel methods hold promise for using existing data to generate additional insights. Objective To evaluate whether contemporary machine learning methods can facilitate risk prediction by including a larger number of variables and identifying complex relationships between predictors and outcomes. Design, Setting, and Participants This cohort study used the American College of Cardiology Chest Pain-MI Registry to identify all AMI hospitalizations between January 1, 2011, and December 31, 2016. Data analysis was performed from February 1, 2018, to October 22, 2020. Main Outcomes and Measures Three machine learning models were developed and validated to predict in-hospital mortality based on patient comorbidities, medical history, presentation characteristics, and initial laboratory values. Models were developed based on extreme gradient descent boosting (XGBoost, an interpretable model), a neural network, and a meta-classifier model. Their accuracy was compared against the current standard developed using a logistic regression model in a validation sample. Results A total of 755 402 patients (mean [SD] age, 65 [13] years; 495 202 [65.5%] male) were identified during the study period. In independent validation, 2 machine learning models, gradient descent boosting and meta-classifier (combination including inputs from gradient descent boosting and a neural network), marginally improved discrimination compared with logistic regression (C statistic, 0.90 for best performing machine learning model vs 0.89 for logistic regression). Nearly perfect calibration in independent validation data was found in the XGBoost (slope of predicted to observed events, 1.01; 95% CI, 0.99-1.04) and the meta-classifier model (slope of predicted-to-observed events, 1.01; 95% CI, 0.99-1.02), with more precise classification across the risk spectrum. The XGBoost model reclassified 32 393 of 121 839 individuals (27%) and the meta-classifier model reclassified 30 836 of 121 839 individuals (25%) deemed at moderate to high risk for death in logistic regression as low risk, which were more consistent with the observed event rates. Conclusions and Relevance In this cohort study using a large national registry, none of the tested machine learning models were associated with substantive improvement in the discrimination of in-hospital mortality after AMI, limiting their clinical utility. However, compared with logistic regression, XGBoost and meta-classifier models, but not the neural network, offered improved resolution of risk for high-risk individuals.