ECG-Based Deep Learning and Clinical Risk Factors to Predict Atrial Fibrillation.

BACKGROUND: Artificial intelligence (AI)-enabled analysis of 12-lead ECGs may facilitate efficient estimation of incident atrial fibrillation (AF) risk. However, it remains unclear whether AI provides meaningful and generalizable improvement in predictive accuracy beyond clinical risk factors for AF. METHODS: We trained a convolutional neural network (ECG-AI) to infer 5-year incident AF risk using 12-lead ECGs in patients receiving longitudinal primary care at Massachusetts General Hospital (MGH). We then fit 3 Cox proportional hazards models, composed of ECG-AI 5-year AF probability, CHARGE-AF clinical risk score (Cohorts for Heart and Aging in Genomic Epidemiology-Atrial Fibrillation), and terms for both ECG-AI and CHARGE-AF (CH-AI), respectively. We assessed model performance by calculating discrimination (area under the receiver operating characteristic curve) and calibration in an internal test set and 2 external test sets (Brigham and Women's Hospital [BWH] and UK Biobank). Models were recalibrated to estimate 2-year AF risk in the UK Biobank given limited available follow-up. We used saliency mapping to identify ECG features most influential on ECG-AI risk predictions and assessed correlation between ECG-AI and CHARGE-AF linear predictors. RESULTS: The training set comprised 45 770 individuals (age 55±17 years, 53% women, 2171 AF events) and the test sets comprised 83 162 individuals (age 59±13 years, 56% women, 2424 AF events). Area under the receiver operating characteristic curve was comparable using CHARGE-AF (MGH, 0.802 [95% CI, 0.767-0.836]; BWH, 0.752 [95% CI, 0.741-0.763]; UK Biobank, 0.732 [95% CI, 0.704-0.759]) and ECG-AI (MGH, 0.823 [95% CI, 0.790-0.856]; BWH, 0.747 [95% CI, 0.736-0.759]; UK Biobank, 0.705 [95% CI, 0.673-0.737]). Area under the receiver operating characteristic curve was highest using CH-AI (MGH, 0.838 [95% CI, 0.807 to 0.869]; BWH, 0.777 [95% CI, 0.766 to 0.788]; UK Biobank, 0.746 [95% CI, 0.716 to 0.776]). Calibration error was low using ECG-AI (MGH, 0.0212; BWH, 0.0129; UK Biobank, 0.0035) and CH-AI (MGH, 0.012; BWH, 0.0108; UK Biobank, 0.0001). In saliency analyses, the ECG P-wave had the greatest influence on AI model predictions. ECG-AI and CHARGE-AF linear predictors were correlated (Pearson r: MGH, 0.61; BWH, 0.66; UK Biobank, 0.41). CONCLUSIONS: AI-based analysis of 12-lead ECGs has similar predictive usefulness to a clinical risk factor model for incident AF and the approaches are complementary. ECG-AI may enable efficient quantification of future AF risk.

Automated Electronic Phenotyping of Cardioembolic Stroke.

BACKGROUND AND PURPOSE: Oral anticoagulation is generally indicated for cardioembolic strokes, but not for other stroke causes. Consequently, subtype classification of ischemic stroke is important for risk stratification and secondary prevention. Because manual classification of ischemic stroke is time-intensive, we assessed the accuracy of automated algorithms for performing cardioembolic stroke subtyping using an electronic health record (EHR) database. METHODS: We adapted TOAST (Trial of ORG 10172 in Acute Stroke Treatment) features associated with cardioembolic stroke for derivation in the EHR. Using administrative codes and echocardiographic reports within Mass General Brigham Biobank (N=13 079), we iteratively developed EHR-based algorithms to define the TOAST cardioembolic stroke features, revising regular expression algorithms until achieving positive predictive value ≥80%. We compared several machine learning-based statistical algorithms for discriminating cardioembolic stroke using the feature algorithms applied to EHR data from 1598 patients with acute ischemic strokes from the Massachusetts General Hospital Ischemic Stroke Registry (2002-2010) with previously adjudicated TOAST and Causative Classification of Stroke subtypes. RESULTS: Regular expression-based feature extraction algorithms achieved a mean positive predictive value of 95% (range, 88%-100%) across 11 echocardiographic features. Among 1598 patients from the Massachusetts General Hospital Ischemic Stroke Registry, 1068 had any cardioembolic stroke feature within predefined time windows in proximity to the stroke event. Cardioembolic stroke tended to occur at an older age, with more TOAST-based comorbidities, and with atrial fibrillation (82.3%). The best model was a random forest with 92.2% accuracy and area under the receiver operating characteristic curve of 91.1% (95% CI, 87.5%-93.9%). Atrial fibrillation, age, dilated cardiomyopathy, congestive heart failure, patent foramen ovale, mitral annulus calcification, and recent myocardial infarction were the most discriminatory features. CONCLUSIONS: Machine learning-based identification of cardioembolic stroke using EHR data is feasible. Future work is needed to improve the accuracy of automated cardioembolic stroke identification and assess generalizability of electronic phenotyping algorithms across clinical settings.

Cohort design and natural language processing to reduce bias in electronic health records research.

Electronic health record (EHR) datasets are statistically powerful but are subject to ascertainment bias and missingness. Using the Mass General Brigham multi-institutional EHR, we approximated a community-based cohort by sampling patients receiving longitudinal primary care between 2001-2018 (Community Care Cohort Project [C3PO], n = 520,868). We utilized natural language processing (NLP) to recover vital signs from unstructured notes. We assessed the validity of C3PO by deploying established risk models for myocardial infarction/stroke and atrial fibrillation. We then compared C3PO to Convenience Samples including all individuals from the same EHR with complete data, but without a longitudinal primary care requirement. NLP reduced the missingness of vital signs by 31%. NLP-recovered vital signs were highly correlated with values derived from structured fields (Pearson r range 0.95-0.99). Atrial fibrillation and myocardial infarction/stroke incidence were lower and risk models were better calibrated in C3PO as opposed to the Convenience Samples (calibration error range for myocardial infarction/stroke: 0.012-0.030 in C3PO vs. 0.028-0.046 in Convenience Samples; calibration error for atrial fibrillation 0.028 in C3PO vs. 0.036 in Convenience Samples). Sampling patients receiving regular primary care and using NLP to recover missing data may reduce bias and maximize generalizability of EHR research.

Predicting colorectal polyp recurrence using time-to-event analysis of medical records.

Identifying patient characteristics that influence the rate of colorectal polyp recurrence can provide important insights into which patients are at higher risk for recurrence. We used natural language processing to extract polyp morphological characteristics from 953 polyp-presenting patients' electronic medical records. We used subsequent colonoscopy reports to examine how the time to polyp recurrence (731 patients experienced recurrence) is influenced by these characteristics as well as anthropometric features using Kaplan-Meier curves, Cox proportional hazards modeling, and random survival forest models. We found that the rate of recurrence differed significantly by polyp size, number, and location and patient smoking status. Additionally, right-sided colon polyps increased recurrence risk by 30% compared to left-sided polyps. History of tobacco use increased polyp recurrence risk by 20% compared to never-users. A random survival forest model showed an AUC of 0.65 and identified several other predictive variables, which can inform development of personalized polyp surveillance plans.

Characterizing the heterogeneity in 5-aminolevulinic acid-induced fluorescence in glioblastoma.

OBJECTIVE: 5-aminolevulinic acid (5-ALA)-induced protoporphyrin IX (PpIX) fluorescence is an effective surgical adjunct for the intraoperative identification of tumor tissue during resection of high-grade gliomas. The use of 5-ALA-induced PpIX fluorescence in glioblastoma (GBM) has been shown to double the extent of gross-total resection and 6-month progression-free survival. The heterogeneity of 5-ALA-induced PpIX fluorescence observed during surgery presents a technical and diagnostic challenge when utilizing this tool intraoperatively. While some regions show bright fluorescence after 5-ALA administration, other regions do not, despite that both regions of the tumor may be histopathologically indistinguishable. The authors examined the biological basis of this heterogeneity using computational methods. METHODS: The authors collected both fluorescent and nonfluorescent GBM specimens from a total of 14 patients undergoing surgery and examined their gene expression profiles. RESULTS: In this study, the authors found that the gene expression patterns characterizing fluorescent and nonfluorescent GBM surgical specimens were profoundly different and were associated with distinct cellular functions and different biological pathways. Nonfluorescent tumor tissue tended to resemble the neural subtype of GBM; meanwhile, fluorescent tumor tissue did not exhibit a prominent pattern corresponding to known subtypes of GBM. Consistent with this observation, neural GBM samples from The Cancer Genome Atlas database exhibited a significantly lower fluorescence score than nonneural GBM samples as determined by a fluorescence gene signature developed by the authors. CONCLUSIONS: These results provide a greater understanding regarding the biological basis of differential fluorescence observed intraoperatively and can provide a basis to identify novel strategies to maximize the effectiveness of fluorescence agents.

Functional network community detection can disaggregate and filter multiple underlying pathways in enrichment analyses.

Differential expression experiments or other analyses often end in a list of genes. Pathway enrichment analysis is one method to discern important biological signals and patterns from noisy expression data. However, pathway enrichment analysis may perform suboptimally in situations where there are multiple implicated pathways - such as in the case of genes that define subtypes of complex diseases. Our simulation study shows that in this setting, standard overrepresentation analysis identifies many false positive pathways along with the true positives. These false positives hamper investigators' attempts to glean biological insights from enrichment analysis. We develop and evaluate an approach that combines community detection over functional networks with pathway enrichment to reduce false positives. Our simulation study demonstrates that a large reduction in false positives can be obtained with a small decrease in power. Though we hypothesized that multiple communities might underlie previously described subtypes of high-grade serous ovarian cancer and applied this approach, our results do not support this hypothesis. In summary, applying community detection before enrichment analysis may ease interpretation for complex gene sets that represent multiple distinct pathways.

Cholinergic control of male mating behavior in hamsters: effects of central oxotremorine treatment.

The responses of rats to intracranial injections of cholinergic drugs implicate acetylcholine in the control of male mating behavior and suggest specific brain areas as mediators of these effects. In particular, past work has linked the medial preoptic area (MPOA) to the control of intromission frequency but implicated areas near the lateral ventricles in effects on the initiation and spacing of intromissions. Studies of responses to systemic cholinergic treatments suggest that acetylcholine is even more important for the control of mating behavior in male hamsters but provide no information on the relevant brain areas. To fill this gap, we observed the effects of central injections of the cholinergic agonist oxotremorine that approached the MPOA along contrasting paths. Both studies suggest that increased cholinergic activity in or near the MPOA can facilitate behavior by reducing the postejaculatory interval and possibly affecting other parts of the mechanisms controlling the initiation of copulation and the efficiency of performance early in an encounter. In addition, oxotremorine caused other changes in behavior that could not be tied to the MPOA and may reflect actions at more dorsal sites, possibly including the bed nucleus of the stria terminalis and medial septum. These effects were notably heterogeneous, including facilitatory and disruptive effects on male behavior along with a facilitation of lordosis responses to manual stimulation. These results emphasize the number and diversity of elements of sexual behavior in hamsters that are under the partial control of forebrain cholinergic mechanisms.