Automatic Classification of Tumor Response From Radiology Reports With Rule-Based Natural Language Processing Integrated Into the Clinical Oncology Workflow.

PURPOSE: Imaging reports in oncology provide critical information about the disease evolution that should be timely shared to tailor the clinical decision making and care coordination of patients with advanced cancer. However, tumor response stays unstructured in free-text and underexploited. Natural language processing (NLP) methods can help provide this critical information into the electronic health records (EHR) in real time to assist health care workers. METHODS: A rule-based algorithm was developed using SAS tools to automatically extract and categorize tumor response within progression or no progression categories. 2,970 magnetic resonance imaging, computed tomography scan, and positron emission tomography French reports were extracted from the EHR of a large comprehensive cancer center to build a 2,637-document training set and a 603-document validation set. The model was also tested on 189 imaging reports from 46 different radiology centers. A tumor dashboard was created in the EHR using the Timeline tool of the vis.js javascript library. RESULTS: An NLP methodology was applied to create an ontology of radiographic terms defining tumor response, mapping text to five main concepts, and application decision rules on the basis of clinical practice RECIST guidelines. The model achieved an overall accuracy of 0.88 (ranging from 0.87 to 0.94), with similar performance on both progression and no progression classification. The overall accuracy was 0.82 on reports from different radiology centers. Data were visualized and organized in a dynamic tumor response timeline. This tool was deployed successfully at our institution both retrospectively and prospectively as part of an automatic pipeline to screen reports and classify tumor response in real time for all metastatic patients. CONCLUSION: Our approach provides an NLP-based framework to structure and classify tumor response from the EHR and integrate tumor response classification into the clinical oncology workflow.

Development, implementation and preliminary evaluation of clinical dashboards in a department of anesthesia.

Clinical dashboards summarize indicators of high-volume patient data in a concise, user-friendly visual format. There are few studies of the use of dashboards to improve professional practice in anesthesiology. The objective of the present study was to describe the user-centered development, implementation and preliminary evaluation of clinical dashboards dealing with anesthesia unit management and quality assessment in a French university medical center. User needs and technical requirements were identified in end user interviews and then synthesized. Several representations were then developed (according to good visualization practice) and submitted to end users for appraisal. Lastly, dashboards were implemented and made accessible for everyday use via the medical center's network. After a period of use, end user feedback on the dashboard platform was collected as a system usability score (range 0 to 100). Seventeen themes (corresponding to 29 questions and 42 indicators) were identified. After prioritization and feasibility assessment, 10 dashboards were ultimately implemented and deployed. The dashboards variously addressed the unit's overall activity, compliance with guidelines on intraoperative hemodynamics, ventilation and monitoring, and documentation of the anesthesia procedure. The mean (standard deviation) system usability score was 82.6 (11.5), which corresponded to excellent usability. We developed clinical dashboards for a university medical center's anesthesia units. The dashboards' deployment was well received by the center's anesthesiologists. The dashboards' impact on activity and practice after several months of use will now have to be assessed.

Automatic Exploitation of YouTube Data: A Study of Videos Published by a French YouTuber During COVID-19 Quarantine in France.

The objective of this study was to test the feasibility of automatically extracting and exploiting data from the YouTube platform, with a focus on the videos produced by the French YouTuber HugoDécrypte during COVID-19 quarantine in France. For this, we used the YouTube API, which allows the automatic collection of data and meta-data of videos. We have identified the main topics addressed in the comments of the videos and assessed their polarity. Our results provide insights on topics trends over the course of the quarantine and highlight users sentiment towards on-going events. The method can be expanded to large video sets to automatically analyse high amount of user-produced data.

Exploring Patient Path Through Sankey Diagram: A Proof of Concept.

Managers, physicians and researchers need to study patient's path for purposes of management, quality of care and research. We present the proof of concept of the use of a flow diagram, the Sankey diagram, to visualize the trajectory of a population that experienced an event. This representation was tested with two case studies in populations from the anesthesia data warehouse of Lille University Hospital. For the 551 patients undergoing a pancreaticoduodenectomy, Sankey diagram helped us identify atypical care paths of patient being transferred too late in an intensive care unit. For 473953 patients who have had anesthesia procedure, Sankey diagram highlighted that mortality and re-operation rates increase with the number of operations. This preliminary work has been well received by end-users and allowed managers, physicians and researchers to visualize the paths of patients and to provide visualization support for research questions. This work will be followed by generalization.