Telemedicine Use for Oncology Clinic Visits at a Large Medical Center During the Onset of COVID-19.

Introduction: The COVID-19 pandemic created an unprecedented surge toward telemedicine. This project investigated oncology clinic telemedicine utilization across the Vanderbilt University Medical Center (VUMC) from January to October 2020. Poorer prognosis and care of oncology patients is expected to be associated with increased emergency department (ED) visits. Methods: January to October 2020 clinic visits were identified from the VUMC's Electronic Data Warehouse (EDW). Oncology patients were identified by ICD-10 code and their EDW ED visit data were extracted. Joinpoint piecewise linear regression evaluated trends in tele-oncology visits. VUMC ED visits were compared for patients who did versus did not use telemedicine for oncology clinic visits. A Welch's two-tailed t-test investigated differences in ED visits/patient between these cohorts (α < 0.05). Results: A sharp increase in tele-oncology clinic visits from January to April 2020 (Monthly Percent Change = 396.26%) was followed by a steady decrease from April to October 2020 (Monthly Percent Change = -20.93%). The difference between these two trends was significant (p < 0.002). Of 18 cancer sites, breast cancers had the highest proportion (29.04%) of tele-oncology visits. There was no significant difference in January to October 2020 ED usage for oncology patients who did (0.40 ED visits/patient) versus did not (0.38 ED visits/patient) utilize telemedicine (p = 0.69). A total of 9.64% of oncology clinic visits from January to October 2020 were telemedicine visits, just below the 13.0% institutional average. Discussion: At the VUMC, tele-oncology spiked in March and April 2020 before decreasing from April to October 2020. Breast cancer clinics were most likely to use tele-oncology. Telemedicine use was not associated with increased ED visits for oncology patients, suggesting telemedicine as an alternative for routine oncology clinics. Oncology clinic telemedicine usage was 18th-highest among 33 specialties at our institutions, and among the lowest of nonsurgical specialties.

Education for Pediatric Gastroenterology Pathology Reports Increases Understanding Ahead of 21st Century Cures Act Rollout.

As a result of the 21st Century Cures Act, Pediatric Gastroenterology (Peds GI) patients and caregivers are receiving pathology reports directly on their online patient portal. This quality improvement project attempted to validate an educational handout about Peds GI pathology reports and identify areas for caregiver education. Methods: Caregivers (n = 59) of patients undergoing Peds GI procedures were surveyed about their knowledge of pathology reports. Caregivers' responses to each of 4 questions were assessed before and after reading an educational handout. These 4 questions questioned the contents and creators of pathology reports, the role of pathologists, and caregiver confidence in interpreting pathology reports. Results: Nearly one-third of caregivers did not know the role of pathologists before education, and one-fifth of caregivers did not know the contents of a pathology report. Caregivers with less than a college degree (n = 21, P = 0.0089), and caregivers of patients undergoing their first procedure (n = 27, P = 0.0022) showed a significant improvement in responses after the educational handout. Conclusions: Educational handouts can increase caregiver understanding of pathology reports, especially for those with lower education levels or those with children undergoing their first procedure.

Sorting Five Human Tumor Types Reveals Specific Biomarkers and Background Classification Genes.

We applied two state-of-the-art, knowledge independent data-mining methods - Dynamic Quantum Clustering (DQC) and t-Distributed Stochastic Neighbor Embedding (t-SNE) - to data from The Cancer Genome Atlas (TCGA). We showed that the RNA expression patterns for a mixture of 2,016 samples from five tumor types can sort the tumors into groups enriched for relevant annotations including tumor type, gender, tumor stage, and ethnicity. DQC feature selection analysis discovered 48 core biomarker transcripts that clustered tumors by tumor type. When these transcripts were removed, the geometry of tumor relationships changed, but it was still possible to classify the tumors using the RNA expression profiles of the remaining transcripts. We continued to remove the top biomarkers for several iterations and performed cluster analysis. Even though the most informative transcripts were removed from the cluster analysis, the sorting ability of remaining transcripts remained strong after each iteration. Further, in some iterations we detected a repeating pattern of biological function that wasn't detectable with the core biomarker transcripts present. This suggests the existence of a "background classification" potential in which the pattern of gene expression after continued removal of "biomarker" transcripts could still classify tumors in agreement with the tumor type.

Discovery and validation of a glioblastoma co-expressed gene module.

Tumors exhibit complex patterns of aberrant gene expression. Using a knowledge-independent, noise-reducing gene co-expression network construction software called KINC, we created multiple RNAseq-based gene co-expression networks relevant to brain and glioblastoma biology. In this report, we describe the discovery and validation of a glioblastoma-specific gene module that contains 22 co-expressed genes. The genes are upregulated in glioblastoma relative to normal brain and lower grade glioma samples; they are also hypo-methylated in glioblastoma relative to lower grade glioma tumors. Among the proneural, neural, mesenchymal, and classical glioblastoma subtypes, these genes are most-highly expressed in the mesenchymal subtype. Furthermore, high expression of these genes is associated with decreased survival across each glioblastoma subtype. These genes are of interest to glioblastoma biology and our gene interaction discovery and validation workflow can be used to discover and validate co-expressed gene modules derived from any co-expression network.

Discovering Condition-Specific Gene Co-Expression Patterns Using Gaussian Mixture Models: A Cancer Case Study.

A gene co-expression network (GCN) describes associations between genes and points to genetic coordination of biochemical pathways. However, genetic correlations in a GCN are only detectable if they are present in the sampled conditions. With the increasing quantity of gene expression samples available in public repositories, there is greater potential for discovery of genetic correlations from a variety of biologically interesting conditions. However, even if gene correlations are present, their discovery can be masked by noise. Noise is introduced from natural variation (intrinsic and extrinsic), systematic variation (caused by sample measurement protocols and instruments), and algorithmic and statistical variation created by selection of data processing tools. A variety of published studies, approaches and methods attempt to address each of these contributions of variation to reduce noise. Here we describe an approach using Gaussian Mixture Models (GMMs) to address natural extrinsic (condition-specific) variation during network construction from mixed input conditions. To demonstrate utility, we build and analyze a condition-annotated GCN from a compendium of 2,016 mixed gene expression data sets from five tumor subtypes obtained from The Cancer Genome Atlas. Our results show that GMMs help discover tumor subtype specific gene co-expression patterns (modules) that are significantly enriched for clinical attributes.