Use of rapid tests for diagnosis of COVID-19 is now commonplace, but questions remain regarding their performance characteristics compared with those of polymerase chain reaction testing. The role ... Visual Rapid Antigen Test Performance for Delta vs. Omicron SARS-CoV-2 Variants. Use of rapid tests for diagnosis of COVID-19 is now commonplace, but questions remain regarding their perfo... Background: It is important to document the performance of rapid antigen tests (Ag-RDTs) in detecting SARS-CoV-2 variants. Objective: To compare the performance of Ag-RDTs in detecting the Delta (B...

Background: Informed consent forms (ICFs) for clinical trials have become increasingly complex, often hindering participant comprehension and engagement due to legal jargon and lengthy content. The recent advances in large language models (LLMs) present an opportunity to streamline the ICF creation process while improving readability, understandability, and actionability. Objectives: This study aims to evaluate the performance of the Mistral 8x22B LLM in generating ICFs with improved readability, understandability, and actionability. Specifically, we evaluate the model’s effectiveness in generating ICFs that are readable, understandable, and actionable while maintaining the accuracy and completeness. Methods: We processed 4 clinical trial protocols from the institutional review board of UMass Chan Medical School using the Mistral 8x22B model to generate key information sections of ICFs. A multidisciplinary team of 8 evaluators, including clinical researchers and health informaticians, assessed the generated ICFs against human-generated counterparts for completeness, accuracy, readability, understandability, and actionability. Readability, Understandability, and Actionability of Key Information indicators, which include 18 binary-scored items, were used to evaluate these aspects, with higher scores indicating greater accessibility, comprehensibility, and actionability of the information. Statistical analysis, including Wilcoxon rank sum tests and intraclass correlation coefficient calculations, was used to compare outputs. Results: LLM-generated ICFs demonstrated comparable performance to human-generated versions across key sections, with no significant differences in accuracy and completeness (P>.10). The LLM outperformed human-generated ICFs in readability (Readability, Understandability, and Actionability of Key Information score of 76.39% vs 66.67%; Flesch-Kincaid grade level of 7.95 vs 8.38) and understandability (90.63% vs 67.19%; P=.02). The LLM-generated content achieved a perfect score in actionability compared with the human-generated version (100% vs 0%; P<.001). Intraclass correlation coefficient for evaluator consistency was high at 0.83 (95% CI 0.64-1.03), indicating good reliability across assessments. Conclusions: The Mistral 8x22B LLM showed promising capabilities in enhancing the readability, understandability, and actionability of ICFs without sacrificing accuracy or completeness. LLMs present a scalable, efficient solution for ICF generation, potentially enhancing participant comprehension and consent in clinical trials.

Telemedicine and commercial wearable devices capable of detecting atrial fibrillation (AF) have revolutionized arrhythmia care during coronavirus disease 2019. However, not much is known about virtual patient-provider interactions or device sharing behaviors. The purpose of this study was to characterize how participants with or at risk of AF are engaging with their providers in the context of telemedicine and using commercially wearable devices to manage their health. We developed a survey to describe participant behaviors around telemedicine encounters and commercial wearable device use. The survey was distributed to participants diagnosed with AF or those at risk of AF (as determined by being at least 65 years old and having a CHA₂DS₂-VASc stroke risk score of >2) in the University of Massachusetts Memorial Health Care system. The survey was distributed to 23,530 patients, and there were 1222 (5.19%) participant responses. Among the participants, 327 (26.8%) had AF and 895 (73.2%) were at risk of AF. Neither device ownership nor device type use differed by AF status. After adjusting for covariates that may influence surveyed participant communication patterns, we found that participants with AF were more likely to share their wearable device-derived data with providers (adjusted odds ratio 1.87; 95% confidence interval 1.02–3.41). Rates of sharing physical activity or sleep data were low for both groups and did not differ by AF status. Compared with participants at risk of developing AF, those with AF were more likely to share heart rate and rhythm data from their commercial wearable devices with providers. However, both groups had similar rates of sharing physical activity and sleep data, telemedicine engagement, and technology use and ownership.

Background: Public health scientists have used spatial tools such as web-based Geographical Information System (GIS) applications to monitor and forecast the progression of the COVID-19 pandemic and track the impact of their interventions. The ability to track SARS-CoV-2 variants and incorporate the social determinants of health with street-level granularity can facilitate the identification of local outbreaks, highlight variant-specific geospatial epidemiology, and inform effective interventions. We developed a novel dashboard, the University of Massachusetts’ Graphical user interface for Geographic Information (MAGGI) variant tracking system that combines GIS, health-associated sociodemographic data, and viral genomic data to visualize the spatiotemporal incidence of SARS-CoV-2 variants with street-level resolution while safeguarding protected health information. The specificity and richness of the dashboard enhance the local understanding of variant introductions and transmissions so that appropriate public health strategies can be devised and evaluated. Objective: We developed a web-based dashboard that simultaneously visualizes the geographic distribution of SARS-CoV-2 variants in Central Massachusetts, the social determinants of health, and vaccination data to support public health efforts to locally mitigate the impact of the COVID-19 pandemic. Methods: MAGGI uses a server-client model–based system, enabling users to access data and visualizations via an encrypted web browser, thus securing patient health information. We integrated data from electronic medical records, SARS-CoV-2 genomic analysis, and public health resources. We developed the following functionalities into MAGGI: spatial and temporal selection capability by zip codes of interest, the detection of variant clusters, and a tool to display variant distribution by the social determinants of health. MAGGI was built on the Environmental Systems Research Institute ecosystem and is readily adaptable to monitor other infectious diseases and their variants in real-time. Results: We created a geo-referenced database and added sociodemographic and viral genomic data to the ArcGIS dashboard that interactively displays Central Massachusetts’ spatiotemporal variants distribution. Genomic epidemiologists and public health officials use MAGGI to show the occurrence of SARS-CoV-2 genomic variants at high geographic resolution and refine the display by selecting a combination of data features such as variant subtype, subject zip codes, or date of COVID-19–positive sample collection. Furthermore, they use it to scale time and space to visualize association patterns between socioeconomics, social vulnerability based on the Centers for Disease Control and Prevention’s social vulnerability index, and vaccination rates. We launched the system at the University of Massachusetts Chan Medical School to support internal research projects starting in March 2021. Conclusions: We developed a COVID-19 variant surveillance dashboard to advance our geospatial technologies to study SARS-CoV-2 variants transmission dynamics. This real-time, GIS-based tool exemplifies how spatial informatics can support public health officials, genomics epidemiologists, infectious disease specialists, and other researchers to track and study the spread patterns of SARS-CoV-2 variants in our communities.

Background: Serial testing for SARS-CoV-2 is recommended to reduce spread of the virus; however, little is known about adherence to recommended testing schedules and reporting practices to health departments. Objective: The Self-Testing for Our Protection from COVID-19 (STOP COVID-19) study aims to examine adherence to a risk-based COVID-19 testing strategy using rapid antigen tests and reporting of test results to health departments. Methods: STOP COVID-19 is a 12-week digital study, facilitated using a smartphone app for testing assistance and reporting. We are recruiting 20,000 participants throughout the United States. Participants are stratified into high- and low-risk groups based on history of COVID-19 infection and vaccination status. High-risk participants are instructed to perform twice-weekly testing for COVID-19 using rapid antigen tests, while low-risk participants test only in the case of symptoms or exposure to COVID-19. All participants complete COVID-19 surveillance surveys, and rapid antigen results are recorded within the smartphone app. Primary outcomes include participant adherence to a risk-based serial testing protocol and percentage of rapid tests reported to health departments. Results: As of February 2022, 3496 participants have enrolled, including 1083 high-risk participants. Out of 13,730 tests completed, participants have reported 13,480 (98.18%, 95% CI 97.9%-98.4%) results to state public health departments with full personal identifying information or anonymously. Among 622 high-risk participants who finished the study period, 35.9% showed high adherence to the study testing protocol. Participants with high adherence reported a higher percentage of test results to the state health department with full identifying information than those in the moderate- or low-adherence groups (high: 71.7%, 95% CI 70.3%-73.1%; moderate: 68.3%, 95% CI 66.0%-70.5%; low: 63.1%, 59.5%-66.6%). Conclusions: Preliminary results from the STOP COVID-19 study provide important insights into rapid antigen test reporting and usage, and can thus inform the use of rapid testing interventions for COVID-19 surveillance.

Background: Surgical site infections (SSIs) are one of the most common health care–associated infections, accounting for nearly 20% of all health care–associated infections in hospitalized patients. SSIs are associated with longer hospital stays, increased readmission rates, higher health care costs, and a mortality rate twice that of patients without infections. Objective: This study aimed to develop and evaluate machine learning (ML) models for augmenting SSI surveillance after colon surgery with the goal of improving the efficiency of infection control practices by prioritizing patients at high risk. Methods: We conducted a retrospective study using data from 1508 patients undergoing colon surgery treated between 2018 and 2023 at a single academic medical center. Of these 1508 patients, 66 (4.4%) developed SSIs as adjudicated by infection control practitioners following Centers for Disease Control and Prevention National Healthcare Safety Network criteria. Data included 78 structured variables (eg, demographics, comorbidities, vital signs, laboratory tests, medications, and operative details) and 2 features derived from unstructured clinical notes using natural language processing. ML models―logistic regression, random forest, and Extreme Gradient Boosting (XGBoost)―were trained using stratified 80/20 train-test splits. Class imbalance was addressed using cost-sensitive learning and the synthetic minority oversampling technique. Model performance was evaluated using precision, recall, F1-score, area under the receiver operating characteristic curve, and Brier scores for calibration. Results: Of the 1508 patients, those who developed SSIs had longer hospital stays (mean 8.1, SD 6.8 days vs mean 6.3, SD 10.5 days; P<.001), higher rates of an American Society of Anesthesiologists score of 3 (52/66, 79% vs 653/1442, 45.3%; P<.001), and elevated white blood cell counts (51/66, 77% vs 734/1442, 50.9%; P<.001). XGBoost achieved the best overall performance with an area under the receiver operating characteristic curve of 0.788, precision of 50%, recall of 38%, and Brier score of 0.035. Random forest yielded perfect precision (100%) but lower recall (23%), with a Brier score of 0.034. Logistic regression showed the highest recall (46%) but the lowest precision (10%), with a Brier score of 0.139. Feature importance analysis using Shapley additive explanations (SHAP) values revealed that the top predictors included recovery duration (SHAP=1.18), SSI keyword frequency (SHAP=1.12), patient age (SHAP=1.12), and American Society of Anesthesiologists score (SHAP=0.94), with natural language processing–derived features ranking among the top 10. Conclusions: ML models can augment traditional SSI surveillance by improving early identification of patients at high risk. The XGBoost model offered the best trade-off between discrimination and calibration, suggesting its utility in clinical workflows. Incorporating structured and unstructured electronic health record data enhances model accuracy and clinical relevance, supporting scalable and efficient infection control practices.