Bayesian Inference for High Dimensional Cox Models with Gaussian and Diffused-Gamma Priors: A Case Study of Mortality in COVID-19 Patients Admitted to the ICU.

Bayesian approaches have been utilized to address the challenge of variable selection and statistical inference in high-dimensional survival analysis. However, the discontinuity of the ℓ0-norm prior, including the useful spike-and-slab prior, may lead to computational and implementation challenges, potentially limiting the widespread use of Bayesian methods. The Gaussian and diffused-gamma (GD) prior has emerged as a promising alternative due to its continuous-and-differentiable ℓ0-norm approximation and computational efficiency in generalized linear models. In this paper, we extend the GD prior to semi-parametric Cox models by proposing a rank-based Bayesian inference procedure with the Cox partial likelihood. We develop a computationally efficient algorithm based on the iterative conditional mode (ICM) and Markov chain Monte Carlo methods for posterior inference. Our simulations demonstrate the effectiveness of the proposed method, and we apply it to an electronic health record dataset to identify risk factors associated with COVID-19 mortality in ICU patients at a regional medical center.

A novel approach to augment single-arm clinical studies with real-world data.

In this paper, we develop a methodology for leveraging real-world data into single-arm clinical trial studies. In recent years, the idea of augmenting randomized clinical trials data with real-world data has emerged as a particularly attractive technique for health organizations and drug developers to accelerate the drug development process. Major regulatory authorities such as the Food and Drug Administration and European Medicines Agency have recognized the potential of utilizing real-world data and are advancing toward making regulatory decisions based on real-world evidence. Several statistical methods have been developed in recent years for borrowing data from real-world sources such as electronic health records, product and disease registries, as well as claims and billing data. We propose a novel approach to augment single-arm clinical trials with the real-world data derived from single or multiple data sources. Furthermore, we illustrate the proposed method in the presence of missing data and conduct simulation studies to evaluate its performance in diverse settings.

Metagenomic Geolocation Prediction Using an Adaptive Ensemble Classifier.

Microbiome samples harvested from urban environments can be informative in predicting the geographic location of unknown samples. The idea that different cities may have geographically disparate microbial signatures can be utilized to predict the geographical location based on city-specific microbiome samples. We implemented this idea first; by utilizing standard bioinformatics procedures to pre-process the raw metagenomics samples provided by the CAMDA organizers. We trained several component classifiers and a robust ensemble classifier with data generated from taxonomy-dependent and taxonomy-free approaches. Also, we implemented class weighting and an optimal oversampling technique to overcome the class imbalance in the primary data. In each instance, we observed that the component classifiers performed differently, whereas the ensemble classifier consistently yielded optimal performance. Finally, we predicted the source cities of mystery samples provided by the organizers. Our results highlight the unreliability of restricting the classification of metagenomic samples to source origins to a single classification algorithm. By combining several component classifiers via the ensemble approach, we obtained classification results that were as good as the best-performing component classifier.

A genomics-informed computational biology platform prospectively predicts treatment responses in AML and MDS patients.

Patients with myelodysplastic syndromes (MDS) or acute myeloid leukemia (AML) are generally older and have more comorbidities. Therefore, identifying personalized treatment options for each patient early and accurately is essential. To address this, we developed a computational biology modeling (CBM) and digital drug simulation platform that relies on somatic gene mutations and gene CNVs found in malignant cells of individual patients. Drug treatment simulations based on unique patient-specific disease networks were used to generate treatment predictions. To evaluate the accuracy of the genomics-informed computational platform, we conducted a pilot prospective clinical study (NCT02435550) enrolling confirmed MDS and AML patients. Blinded to the empirically prescribed treatment regimen for each patient, genomic data from 50 evaluable patients were analyzed by CBM to predict patient-specific treatment responses. CBM accurately predicted treatment responses in 55 of 61 (90%) simulations, with 33 of 61 true positives, 22 of 61 true negatives, 3 of 61 false positives, and 3 of 61 false negatives, resulting in a sensitivity of 94%, a specificity of 88%, and an accuracy of 90%. Laboratory validation further confirmed the accuracy of CBM-predicted activated protein networks in 17 of 19 (89%) samples from 11 patients. Somatic mutations in the TET2, IDH1/2, ASXL1, and EZH2 genes were discovered to be highly informative of MDS response to hypomethylating agents. In sum, analyses of patient cancer genomics using the CBM platform can be used to predict precision treatment responses in MDS and AML patients.

Bayesian disease classification using copy number data.

DNA copy number variations (CNVs) have been shown to be associated with cancer development and progression. The detection of these CNVs has the potential to impact the basic knowledge and treatment of many types of cancers, and can play a role in the discovery and development of molecular-based personalized cancer therapies. One of the most common types of high-resolution chromosomal microarrays is array-based comparative genomic hybridization (aCGH) methods that assay DNA CNVs across the whole genomic landscape in a single experiment. In this article we propose methods to use aCGH profiles to predict disease states. We employ a Bayesian classification model and treat disease states as outcome, and aCGH profiles as covariates in order to identify significant regions of the genome associated with disease subclasses. We propose a principled two-stage method where we first make inferences on the underlying copy number states associated with the aCGH emissions based on hidden Markov model (HMM) formulations to account for serial dependencies in neighboring probes. Subsequently, we infer associations with disease outcomes, conditional on the copy number states, using Bayesian linear variable selection procedures. The selected probes and their effects are parameters that are useful for predicting the disease categories of any additional individuals on the basis of their aCGH profiles. Using simulated datasets, we investigate the method's accuracy in detecting disease category. Our methodology is motivated by and applied to a breast cancer dataset consisting of aCGH profiles assayed on patients from multiple disease subtypes.

Bayesian Hidden Markov Modeling of Array CGH Data.

Genomic alterations have been linked to the development and progression of cancer. The technique of comparative genomic hybridization (CGH) yields data consisting of fluorescence intensity ratios of test and reference DNA samples. The intensity ratios provide information about the number of copies in DNA. Practical issues such as the contamination of tumor cells in tissue specimens and normalization errors necessitate the use of statistics for learning about the genomic alterations from array CGH data. As increasing amounts of array CGH data become available, there is a growing need for automated algorithms for characterizing genomic profiles. Specifically, there is a need for algorithms that can identify gains and losses in the number of copies based on statistical considerations, rather than merely detect trends in the data.We adopt a Bayesian approach, relying on the hidden Markov model to account for the inherent dependence in the intensity ratios. Posterior inferences are made about gains and losses in copy number. Localized amplifications (associated with oncogene mutations) and deletions (associated with mutations of tumor suppressors) are identified using posterior probabilities. Global trends such as extended regions of altered copy number are detected. Because the posterior distribution is analytically intractable, we implement a Metropolis-within-Gibbs algorithm for efficient simulation-based inference. Publicly available data on pancreatic adenocarcinoma, glioblastoma multiforme, and breast cancer are analyzed, and comparisons are made with some widely used algorithms to illustrate the reliability and success of the technique.