Bayesian consensus clustering for multivariate longitudinal data

In clinical and epidemiological studies, there is a growing interest in studying the heterogeneity among patients based on longitudinal characteristics to identify subtypes of the study population. Compared to clustering a single longitudinal marker, simultaneously clustering multiple longitudinal markers allow additional information to be incorporated into the clustering process, which reveals co-existing longitudinal patterns and generates deeper biological insight. In the current study, we propose a Bayesian consensus clustering (BCC) model for multivariate longitudinal data. Instead of arriving at a single overall clustering, the proposed model allows each marker to follow marker-specific local clustering and these local clusterings are aggregated to find a global (consensus) clustering. To estimate the posterior distribution of model parameters, a Gibbs sampling algorithm is proposed. We apply our proposed model to the primary biliary cirrhosis study to identify patient subtypes that may be associated with their prognosis. We also perform simulation studies to compare the clustering performance between the proposed model and existing models under several scenarios. The results demonstrate that the proposed BCC model serves as a useful tool for clustering multivariate longitudinal data.     

Tissue tracking of segmental strain as a predictor of provoked dynamic left ventricular outflow tract obstruction

BackgroundLeft ventricular outflow tract obstruction (LVOTO) is commonly observed in patients with hypertrophic cardiomyopathy (HCM) or left ventricular hypertrophy (LVH). Some patients develop LVOTO provoked by physical exertion, and hence termed dynamic LVOTO (DLVOTO). However, its precise prevalence and mechanism are still unclear.AimTwo-dimensional speckle tracking echocardiography (2D STE) seems to be helpful for the detection of early LV structural abnormalities. This study aimed to examine the possible role of segmental as well as global longitudinal strain in identifying DLVOTO non-HCM patients as detected by dobutamine stress echocardiography (DSE).Methods and resultsTwo hundred and fifty patients without structural heart disease had undergone conventional transthoracic echocardiography, 2D STE, and DSE. All patients with non-ischemic evidence were divided into two groups according to the DSE results; DLVOTO (+) and DLVOTO (?).Among 250 patients, 50 patients (36%) had shown DLVOTO after DSE (15 males, 35 females; mean age 55±7years). They were compared with 90 non -LVOTO obstruction patients (43 males, 47 females; mean age 57±6years). Based on multivariate logistic regression analysis, the independent predictors of provoked DLVOTO during DSE were resting basal septal longitudinal strain BS-LS _average (p < 0.001), resting LA reservoir strain (p < 0.001), and systolic LVOT diameter (p = 0.03). Resting BS-LS _average with cut-off - 17.5% was recognized as a critical indicator of DLVOTO, with sensitivity 78%, and specificity 95% (better than systolic LVOT diameter of sensitivity 76%, and specificity 15% and resting LA reservoir strain which showed poor AUC at ROC curve 0.007).ConclusionWe demonstrate that provoked LVOTO during DSE in non HCM symptomatic patients is directly correlated to resting regional LS, where the increased BS-LS of ≥ ?17.5% was a key determinant of LVOT gradient provocation. Assessment of baseline BS-LS _average might be a bedside simple tool for detection of patients with DLVOTO not able to do DSE.     

The impact of neighborhoods and friendships on interracial anxiety among medical students and residents: A report from the medical student CHANGES study

Objective

To examine the experience of interracial anxiety among health professionals and how it may affect the quality of their interactions with patients from racially marginalized populations. We explored the influence of prior interracial exposure—specifically through childhood neighborhoods, college student bodies, and friend groups—on interracial anxiety among medical students and residents. We also examined whether levels of interracial anxiety change from medical school through residency.

Data Source

Web-based longitudinal survey data from the Medical Student Cognitive Habits and Growth Evaluation Study.

Study Design

We used a retrospective longitudinal design with four observations for each trainee. The study population consisted of non-Black US medical trainees surveyed in their 1st and 4th years of medical school and 2nd and 3rd years of residency. Mixed effects longitudinal models were used to assess predictors of interracial anxiety and assess changes in interracial anxiety scores over time.

Principal Findings

In total, 3155 non-Black medical trainees were followed for 7 years. Seventy-eight percent grew up in predominantly White neighborhoods. Living in predominantly White neighborhoods and having less racially diverse friends were associated with higher levels of interracial anxiety among medical trainees. Trainees' interracial anxiety scores did not substantially change over time; interracial anxiety was highest in the 1st year of medical school, lowest in the 4th year, and increased slightly during residency.

Conclusions

Neighborhood and friend group composition had independent effects on interracial anxiety, indicating that premedical racial socialization may affect medical trainees' preparedness to interact effectively with diverse patient populations. Additionally, the lack of substantial change in interracial anxiety throughout medical training suggests the importance of providing curricular tools and structure (e.g., instituting interracial cooperative learning activities) to foster the development of healthy interracial relationships.