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.     

Validation of PATHFx 2.0: An open-source tool for estimating survival in patients undergoing pathologic fracture fixation

Treatment decisions in patients with metastatic bone disease rely on accurate survival estimation. We developed the original PATHFx models using expensive, proprietary software and now seek to provide a more cost-effective solution. Using open-source machine learning software to create PATHFx version 2.0, we asked whether PATHFx 2.0 could be created using open-source methods and externally validated in two unique patient populations. The training set of a well-characterized, database records of 189 patients and the bnlearn package within R Version 3.5.1 (R Foundation for Statistical Computing), was used to establish a series of Bayesian belief network models designed to predict survival at 1, 3, 6, 12, 18, and 24 months. Each was externally validated in both a Scandinavian (n = 815 patients) and a Japanese (n = 261 patients) data set. Brier scores and receiver operating characteristic curves to assessed discriminatory ability. Decision curve analysis (DCA) evaluated whether models should be used clinically. DCA showed that the model should be used clinically at all time points in the Scandinavian data set. For the 1-month time point, DCA of the Japanese data set suggested to expect better outcomes assuming all patients will survive greater than 1 month. Brier scores for each curve demonstrate that the models are accurate at each time point. Statement of Clinical Significance: we successfully transitioned to PATHFx 2.0 using open-source software and externally validated it in two unique patient populations, which can be used as a cost-effective option to guide surgical decisions in patients with metastatic bone disease.     

Mise à jour des croyances et congruence à l'humeur dans le trouble dépressif

Depressive disorder is characterized by a polymorphic symptomatology associating emotional, cognitive and behavioral disturbances. One of the most specific symptoms is negative beliefs, called congruent to mood. Despite the importance of these beliefs in the development, the maintenance, and the recurrence of depressive episodes, little is known about the processes underlying the generation of depressive beliefs. In this paper, we detail the link between belief updating mechanisms and the genesis of depressive beliefs. We show how depression alters information processing, generating cognitive immunization when processing positive information, affective updating bias related to the valence of belief and prediction error, and difficultie to disengage from negative information. We suggest that disruption of belief-updating mechanisms forms the basis of belief-mood congruence in depression.     

混合Poisson分布及其应用：疾病的统计分布（五）

本文讨论了混合Ｐｏｉｓｓｏｎ分布的性质、应用条件、参数的估计及混合Ｐｏｉｓｓｏｎ分布阶数的确定，指出混合Ｐｏｉｓｓｏｎ分布可用于混合样本的判别归类，并用Ｂａｙｅｓ的思想导出其判别归类方法。模拟试验结果表明：当混合Ｐｏｉｓｓｏｎ分布中各部分的比例相差不大，而各部分的均值相差较大时，抽样效果和拟合效果越好，所得到的估计值越接近理论值；反之效果越差。     

Intelligent dialogue based on statistical models of clinical decision-making

The independence Bayesian model has been used widely in computer programs designed to support clinical decision-making. A reasoning strategy has been developed to enable these programs to conduct clinically pertinent dialogue and explain their reasoning. It has been implemented in a program for the diagnosis of acute abdominal pain based on the Bayesian model of de Dombal et al. Several features of the dialogue design have been adopted from artificial intelligence research, including shared initiative and critiquing. The program adopts a flexible goal-driven strategy, attempting to confirm the clinician's diagnosis or rule out the likeliest alternative. Symptoms and signs are selected in order of their expected weights of evidence in favour of the hypothesized disease.     

Laboratory screening method for selection of healthy volunteers

The aim of laboratory screening in Phase I is to exclude subjects with subclinical illness, who might be at increased risk in the study, and who might also adversely influence interpretation of the results. A new method for laboratory screening, based on Bayesian probability theory, is proposed, which consists of: 1. Drawing up a list of diseases to be excluded. 2. Defining for each disease, the maximum acceptable risk that an included subject could be affected by it. 3. Identifying one test for each disease. 4. Using a contingency table to calculate the specificity of the test and integrating the estimated prevalence of the disease from epidemiological data. 5. Applying the percentage obtained by the calculation of specificity to the previously determined distribution of values in the volunteer population to identify the threshold value for inclusion. Use of this deductive method in screening volunteers for Phase I trials affords increased security of selection, while reducing the number of non-pertinent exclusions because of laboratory findings.     

Real-time refinement of subthalamic nucleus targeting using Bayesian decision-making on the root mean square measure.

The subthalamic nucleus (STN) is a major target for treatment of advanced Parkinson's disease patients undergoing deep brain stimulation surgery. Microelectrode recording (MER) is used in many cases to identify the target nucleus. A real-time procedure for identifying the entry and exit points of the STN would improve the outcome of this targeting procedure. We used the normalized root mean square (NRMS) of a short (5 seconds) MER sampled signal and the estimated anatomical distance to target (EDT) as the basis for this procedure. Electrode tip location was defined intraoperatively by an expert neurophysiologist to be before, within, or after the STN. Data from 46 trajectories of 27 patients were used to calculate the Bayesian posterior probability of being in each of these locations, given RMS-EDT pair values. We tested our predictions on each trajectory using a bootstrapping technique, with the rest of the trajectories serving as a training set and found the error in predicting the STN entry to be (mean +/- SD) 0.18 +/- 0.84, and 0.50 +/- 0.59 mm for STN exit point, which yields a 0.30 +/- 0.28 mm deviation from the expert's target center. The simplicity and computational ease of RMS calculation, its spike sorting-independent nature and tolerance to electrode parameters of this Bayesian predictor, can lead directly to the development of a fully automated intraoperative physiological procedure for the refinement of imaging estimates of STN borders.     

Adaptive group sequential design for phase II clinical trials: A Bayesian decision theoretic approach

Bayesian decision theoretic approaches (BDTAs) have been widely studied in the literature as tools for designing and conducting phase II clinical trials. However, full Bayesian approaches that consider multiple endpoints are lacking. Since the monitoring of toxicity is a major goal of phase II trials, we propose an adaptive group sequential design using a BDTA, which characterizes efficacy and toxicity as correlated bivariate binary endpoints. We allow trade‐off between the two endpoints. Interim evaluations are conducted group sequentially, but the number of interim looks and the size of each group are chosen adaptively based on current observations. We utilize a loss function consisting of two components: the loss associated with accruing, treating, and monitoring patients, and the loss associated with making incorrect decisions. The performance of our Bayesian modeling, and the operating characteristics of decision rules under a wide range of loss function parameters are evaluated using seven scenarios in a simulation study. Our method is illustrated in the context of a single‐arm phase II trial of bevacizumab, gemcitabine, and oxaliplatin in patients with metastatic pancreatic adenocarcinoma. Copyright © 2009 John Wiley & Sons, Ltd.