BACKGROUND: Longitudinal data on key cancer outcomes for clinical research, such as response to treatment and disease progression, are not captured in standard cancer registry reporting. Manual extraction of such outcomes from unstructured electronic health records is a slow, resource-intensive process. Natural language processing (NLP) methods can accelerate outcome annotation, but they require substantial labeled data. Transfer learning based on language modeling, particularly using the Transformer architecture, has achieved improvements in NLP performance. However, there has been no systematic evaluation of NLP model training strategies on the extraction of cancer outcomes from unstructured text. RESULTS: We evaluated the performance of nine NLP models at the two tasks of identifying cancer response and cancer progression within imaging reports at a single academic center among patients with non-small cell lung cancer. We trained the classification models under different conditions, including training sample size, classification architecture, and language model pre-training. The training involved a labeled dataset of 14,218 imaging reports for 1112 patients with lung cancer. A subset of models was based on a pre-trained language model, DFCI-ImagingBERT, created by further pre-training a BERT-based model using an unlabeled dataset of 662,579 reports from 27,483 patients with cancer from our center. A classifier based on our DFCI-ImagingBERT, trained on more than 200 patients, achieved the best results in most experiments; however, these results were marginally better than simpler "bag of words" or convolutional neural network models. CONCLUSION: When developing AI models to extract outcomes from imaging reports for clinical cancer research, if computational resources are plentiful but labeled training data are limited, large language models can be used for zero- or few-shot learning to achieve reasonable performance. When computational resources are more limited but labeled training data are readily available, even simple machine learning architectures can achieve good performance for such tasks.

MOTIVATION: Large-scale sequencing studies have created a need to succinctly visualize genomic characteristics of patient cohorts linked to widely variable phenotypic information. This is often done by visualizing the co-occurrence of variants with comutation plots. Current tools lack the ability to create highly customizable and publication quality comutation plots from arbitrary user data. RESULTS: We developed CoMut, a stand-alone, object-oriented Python package that creates comutation plots from arbitrary input data, including categorical data, continuous data, bar graphs, side bar graphs and data that describes relationships between samples. AVAILABILITY AND IMPLEMENTATION: The CoMut package is open source and is available at https://github.com/vanallenlab/comut under the MIT License, along with documentation and examples. A no installation, easy-to-use implementation is available on Google Colab (see GitHub).

Objective: Misinterpretation of complex genomic data presents a major challenge in the implementation of precision oncology. We sought to determine whether interactive genomic reports with embedded clinician education and optimized data visualization improved genomic data interpretation. Materials and Methods: We conducted a randomized, vignette-based survey study to determine whether exposure to interactive reports for a somatic gene panel, as compared to static reports, improves physicians' genomic comprehension and report-related satisfaction (overall scores calculated across 3 vignettes, range 0-18 and 1-4, respectively, higher score corresponding with improved endpoints). Results: One hundred and five physicians at a tertiary cancer center participated (29% participation rate): 67% medical, 20% pediatric, 7% radiation, and 7% surgical oncology; 37% female. Prior to viewing the case-based vignettes, 34% of the physicians reported difficulty making treatment recommendations based on the standard static report. After vignette/report exposure, physicians' overall comprehension scores did not differ by report type (mean score: interactive 11.6 vs static 10.5, difference = 1.1, 95% CI, -0.3, 2.5, P = .13). However, physicians exposed to the interactive report were more likely to correctly assess sequencing quality (P < .001) and understand when reports needed to be interpreted with caution (eg, low tumor purity; P = .02). Overall satisfaction scores were higher in the interactive group (mean score 2.5 vs 2.1, difference = 0.4, 95% CI, 0.2-0.7, P = .001). Discussion and Conclusion: Interactive genomic reports may improve physicians' ability to accurately assess genomic data and increase report-related satisfaction. Additional research in users' genomic needs and efforts to integrate interactive reports into electronic health records may facilitate the implementation of precision oncology.

OBJECTIVE: Clinicians' ability to use and interpret genetic information depends upon how those data are displayed in electronic health records (EHRs). There is a critical need to develop systems to effectively display genetic information in EHRs and augment clinical decision support (CDS). MATERIALS AND METHODS: The National Institutes of Health (NIH)-sponsored Clinical Sequencing Exploratory Research and Electronic Medical Records & Genomics EHR Working Groups conducted a multiphase, iterative process involving working group discussions and 2 surveys in order to determine how genetic and genomic information are currently displayed in EHRs, envision optimal uses for different types of genetic or genomic information, and prioritize areas for EHR improvement. RESULTS: There is substantial heterogeneity in how genetic information enters and is documented in EHR systems. Most institutions indicated that genetic information was displayed in multiple locations in their EHRs. Among surveyed institutions, genetic information enters the EHR through multiple laboratory sources and through clinician notes. For laboratory-based data, the source laboratory was the main determinant of the location of genetic information in the EHR. The highest priority recommendation was to address the need to implement CDS mechanisms and content for decision support for medically actionable genetic information. CONCLUSION: Heterogeneity of genetic information flow and importance of source laboratory, rather than clinical content, as a determinant of information representation are major barriers to using genetic information optimally in patient care. Greater effort to develop interoperable systems to receive and consistently display genetic and/or genomic information and alert clinicians to genomic-dependent improvements to clinical care is recommended.