Identifying Electronic Nicotine Delivery System Brands and Flavors on Instagram: Natural Language Processing Analysis.

BACKGROUND: Electronic nicotine delivery system (ENDS) brands, such as JUUL, used social media as a key component of their marketing strategy, which led to massive sales growth from 2015 to 2018. During this time, ENDS use rapidly increased among youths and young adults, with flavored products being particularly popular among these groups. OBJECTIVE: The aim of our study is to develop a named entity recognition (NER) model to identify potential emerging vaping brands and flavors from Instagram post text. NER is a natural language processing task for identifying specific types of words (entities) in text based on the characteristics of the entity and surrounding words. METHODS: NER models were trained on a labeled data set of 2272 Instagram posts coded for ENDS brands and flavors. We compared three types of NER models-conditional random fields, a residual convolutional neural network, and a fine-tuned distilled bidirectional encoder representations from transformers (FTDB) network-to identify brands and flavors in Instagram posts with key model outcomes of precision, recall, and F1 scores. We used data from Nielsen scanner sales and Wikipedia to create benchmark dictionaries to determine whether brands from established ENDS brand and flavor lists were mentioned in the Instagram posts in our sample. To prevent overfitting, we performed 5-fold cross-validation and reported the mean and SD of the model validation metrics across the folds. RESULTS: For brands, the residual convolutional neural network exhibited the highest mean precision (0.797, SD 0.084), and the FTDB exhibited the highest mean recall (0.869, SD 0.103). For flavors, the FTDB exhibited both the highest mean precision (0.860, SD 0.055) and recall (0.801, SD 0.091). All NER models outperformed the benchmark brand and flavor dictionary look-ups on mean precision, recall, and F1. Comparing between the benchmark brand lists, the larger Wikipedia list outperformed the Nielsen list in both precision and recall. CONCLUSIONS: Our findings suggest that NER models correctly identified ENDS brands and flavors in Instagram posts at rates competitive with, or better than, others in the published literature. Brands identified during manual annotation showed little overlap with those in Nielsen scanner data, suggesting that NER models may capture emerging brands with limited sales and distribution. NER models address the challenges of manual brand identification and can be used to support future infodemiology and infoveillance studies. Brands identified on social media should be cross-validated with Nielsen and other data sources to differentiate emerging brands that have become established from those with limited sales and distribution.

Subgroups of failure after surgery for pelvic organ prolapse and associations with quality of life outcomes: a longitudinal cluster analysis.

BACKGROUND: Treatment outcomes after pelvic organ prolapse surgery are often presented as dichotomous "success or failure" based on anatomic and symptom criteria. However, clinical experience suggests that some women with outcome "failures" are asymptomatic and perceive their surgery to be successful and that other women have anatomic resolution but continue to report symptoms. Characterizing failure types could be a useful step to clarify definitions of success, understand mechanisms of failure, and identify individuals who may benefit from specific therapies. OBJECTIVE: This study aimed to identify clusters of women with similar failure patterns over time and assess associations among clusters and the Pelvic Organ Prolapse Distress Inventory, Short-Form Six-Dimension health index, Patient Global Impression of Improvement, patient satisfaction item questionnaire, and quality-adjusted life-year. STUDY DESIGN: Outcomes were evaluated for up to 5 years in a cohort of participants (N=709) with stage ≥2 pelvic organ prolapse who underwent surgical pelvic organ prolapse repair and had sufficient follow-up in 1 of 4 multicenter surgical trials conducted by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Pelvic Floor Disorders Network. Surgical success was defined as a composite measure requiring anatomic success (Pelvic Organ Prolapse Quantification system points Ba, Bp, and C of ≤0), subjective success (absence of bothersome vaginal bulge symptoms), and absence of retreatment for pelvic organ prolapse. Participants who experienced surgical failure and attended ≥4 visits from baseline to 60 months after surgery were longitudinally clustered, accounting for similar trajectories in Ba, Bp, and C and degree of vaginal bulge bother; moreover, missing data were imputed. Participants with surgical success were grouped into a separate cluster. RESULTS: Surgical failure was reported in 276 of 709 women (39%) included in the analysis. Failures clustered into the following 4 mutually exclusive subgroups: (1) asymptomatic intermittent anterior wall failures, (2) symptomatic intermittent anterior wall failures, (3) asymptomatic intermittent anterior and posterior wall failures, and (4) symptomatic all-compartment failures. Each cluster had different bulge symptoms, anatomy, and retreatment associations with quality of life outcomes. Asymptomatic intermittent anterior wall failures (n=150) were similar to surgical successes with Ba values that averaged around -1 cm but fluctuated between anatomic success (Ba≤0) and failure (Ba>0) over time. Symptomatic intermittent anterior wall failures (n=82) were anatomically similar to asymptomatic intermittent anterior failures, but women in this cluster persistently reported bothersome bulge symptoms and the lowest quality of life, Short-Form Six-Dimension health index scores, and perceived success. Women with asymptomatic intermittent anterior and posterior wall failures (n=28) had the most severe preoperative pelvic organ prolapse but the lowest symptomatic failure rate and retreatment rate. Participants with symptomatic all-compartment failures (n=16) had symptomatic and anatomic failure early after surgery and the highest retreatment of any cluster. CONCLUSION: In particular, the following 4 clusters of pelvic organ prolapse surgical failure were identified in participants up to 5 years after pelvic organ prolapse surgery: asymptomatic intermittent anterior wall failures, symptomatic intermittent anterior wall failures, asymptomatic intermittent anterior and posterior wall failures, and symptomatic all-compartment failures. These groups provide granularity about the nature of surgical failures after pelvic organ prolapse surgery. Future work is planned for predicting these distinct outcomes using patient characteristics that can be used for counseling women individually.

Supplementing a survey with respondent Twitter data to measure e-cigarette information exposure.

Social media data are increasingly used by researchers to gain insights on individuals' behaviors and opinions. Platforms like Twitter provide access to individuals' postings, networks of friends and followers, and the content to which they are exposed. This article presents the methods and results of an exploratory study to supplement survey data with respondents' Twitter postings, networks of Twitter friends and followers, and information to which they were exposed about e-cigarettes. Twitter use is important to consider in e-cigarette research and other topics influenced by online information sharing and exposure. Further, Twitter metadata provide direct measures of user's friends and followers as opposed to survey self-reports. We find that Twitter metadata provide similar information to survey questions on Twitter network size without inducing recall error or other measurement issues. Using sentiment coding and machine learning methods, we find Twitter can elucidate on topics difficult to measure via surveys such as online expressed opinions and network composition. We present and discuss models predicting whether respondents' tweet positively about e-cigarettes using survey and Twitter data, finding the combined data to provide broader measures than either source alone.

Underutilization and disparities in access to EGFR testing among Medicare patients with lung cancer from 2010 - 2013.

BACKGROUND: Tumor testing for mutations in the epidermal growth factor receptor (EGFR) gene is indicated for all newly diagnosed, metastatic lung cancer patients, who may be candidates for first-line treatment with an EGFR tyrosine kinase inhibitor. Few studies have analyzed population-level testing. METHODS: We identified clinical, demographic, and regional predictors of EGFR & KRAS testing among Medicare beneficiaries with a new diagnosis of lung cancer in 2011-2013 claims. The outcome variable was whether the patient underwent molecular, EGFR and KRAS testing. Independent variables included: patient demographics, Medicaid status, clinical characteristics, and region where the patient lived. We performed multivariate logistic regression to identify factors that predicted testing. RESULTS: From 2011 to 2013, there was a 19.7% increase in the rate of EGFR testing. Patient zip code had the greatest impact on odds to undergo testing; for example, patients who lived in the Boston, Massachusetts hospital referral region were the most likely to be tested (odds ratio (OR) of 4.94, with a 95% confidence interval (CI) of 1.67-14.62). Patient demographics also impacted odds to be tested. Asian/Pacific Islanders were most likely to be tested (OR 1.63, CI 1.53-1.79). Minorities and Medicaid patients were less likely to be tested. Medicaid recipients had an OR of 0.74 (CI 0.72-0.77). Hispanics and Blacks were also less likely to be tested (OR 0.97, CI 0.78-0.99 and 0.95, CI 0.92-0.99), respectively. Clinical procedures were also correlated with testing. Patients who underwent transcatheter biopsies were 2.54 times more likely to be tested (CI 2.49-2.60) than those who did not undergo this type of biopsy. CONCLUSIONS: Despite an overall increase in EGFR testing, there is widespread underutilization of guideline-recommended testing. We observed racial, income, and regional disparities in testing. Precision medicine has increased the complexity of cancer diagnosis and treatment. Targeted interventions and clinical decision support tools are needed to ensure that all patients are benefitting from advances in precision medicine. Without such interventions, precision medicine may exacerbate racial disparities in cancer care and health outcomes.

Insights from an N3C RECOVER EHR-based cohort study characterizing SARS-CoV-2 reinfections and Long COVID.

BACKGROUND: Although the COVID-19 pandemic has persisted for over 3 years, reinfections with SARS-CoV-2 are not well understood. We aim to characterize reinfection, understand development of Long COVID after reinfection, and compare severity of reinfection with initial infection. METHODS: We use an electronic health record study cohort of over 3 million patients from the National COVID Cohort Collaborative as part of the NIH Researching COVID to Enhance Recovery Initiative. We calculate summary statistics, effect sizes, and Kaplan-Meier curves to better understand COVID-19 reinfections. RESULTS: Here we validate previous findings of reinfection incidence (6.9%), the occurrence of most reinfections during the Omicron epoch, and evidence of multiple reinfections. We present findings that the proportion of Long COVID diagnoses is higher following initial infection than reinfection for infections in the same epoch. We report lower albumin levels leading up to reinfection and a statistically significant association of severity between initial infection and reinfection (chi-squared value: 25,697, p-value: <0.0001) with a medium effect size (Cramer's V: 0.20, DoF = 3). Individuals who experienced severe initial and first reinfection were older in age and at a higher mortality risk than those who had mild initial infection and reinfection. CONCLUSIONS: In a large patient cohort, we find that the severity of reinfection appears to be associated with the severity of initial infection and that Long COVID diagnoses appear to occur more often following initial infection than reinfection in the same epoch. Future research may build on these findings to better understand COVID-19 reinfections.

More than three years after the start of the COVID-19 pandemic, individuals are frequently reporting multiple COVID-19 infections. However, these reinfections remain poorly understood. Here, we investigate COVID-19 reinfections in a large electronic health record cohort of over 3 million patients. We use data summary techniques and statistical tests to characterize reinfections and their relationships with disease severity, biomarkers, and Long COVID. We find that individuals with severe initial infection are more likely to experience severe reinfection, that some protein levels are lower, leading to reinfection, and that a lower proportion of individuals are diagnosed with Long COVID following reinfection than initial infection. Our work highlights the prevalence and impact of reinfections and suggests the need for further research.

SARS-CoV-2 Reinfection is Preceded by Unique Biomarkers and Related to Initial Infection Timing and Severity: an N3C RECOVER EHR-Based Cohort Study.

Although the COVID-19 pandemic has persisted for over 2 years, reinfections with SARS-CoV-2 are not well understood. We use the electronic health record (EHR)-based study cohort from the National COVID Cohort Collaborative (N3C) as part of the NIH Researching COVID to Enhance Recovery (RECOVER) Initiative to characterize reinfection, understand development of Long COVID after reinfection, and compare severity of reinfection with initial infection. We validate previous findings of reinfection incidence (5.9%), the occurrence of most reinfections during the Omicron epoch, and evidence of multiple reinfections. We present novel findings that Long COVID diagnoses occur closer to the index date for infection or reinfection in the Omicron BA epoch. We report lower albumin levels leading up to reinfection and a statistically significant association of severity between first infection and reinfection (chi-squared value: 9446.2, p-value: 0) with a medium effect size (Cramer's V: 0.18, DoF = 4).