Physician-Led Initiative to Diagnostic Imaging Delayed Results Goes Live in November
July 28, 2020 | 2 min. to read
Northwestern Medicine has embarked on a journey to improve identification and tracking of important findings on diagnostic images. National and local analyses have highlighted the need for better systems to ensure follow up of important findings on radiology images, especially for lung nodules. After in-depth analysis of our experience at NM, as well as national evaluation of best practices, a new process has been developed to better identify, communicate and track the follow-up of X-ray findings.
In collaboration with Northwestern University, artificial intelligence solutions were developed and will be deployed in the months ahead. Natural language processing tools were developed that will review dictated reports to identify the finding, and recognize the recommend follow-up and time frame. The tool will generate an InBasket message in Epic for the ordering physician or other appropriate physician. The InBasket message will clearly and simply display the completed imaging study result and follow-up recommendation.
Physicians will be able to automatically order the recommended follow-up, cancel if it is not clinically required or easily transfer responsibility if they are not the appropriate physician. Follow-up orders can be tracked in the system until completed. If a patient does not complete the follow-up in the specified time and/or does not have a primary care physician on record, a team of nurses will reach out to the patient to ensure follow-up is completed and documented in the medical record.
Project work is now focused on build and implementation of the new process with a scheduled go-live of November 2020. The first diagnostic findings that are being built into the new workflow are lung and adrenal nodules. Once the process is rolled out systemwide, additional findings could be added. In order to do so, the feedback of physicians from across the health system will be necessary. The initiative is designed to be simple and supportive, recognizing that physicians are already overloaded with tasks. However, the real-world application of AI tools is new, so an active dialogue will be maintained to ensure that these goals are met.
A team of physicians from across all regions and multiple specialties are currently providing guidance on aspects of the build and implementation. Further communication will be shared in each region when information regarding progress of the build and functionality within the physician workflow is available.
Looking ahead, the project team would welcome your thoughts about the next findings that can be tracked by AI. Work will continue to identify and implement critical findings that can easily be overlooked. The hope is to use AI tracking to enhance relationships, reliability and efficiency for all.