Integration of automation into an existing clinical workflow to improve efficiency and reduce errors in the manual treatment planning process for total body irradiation (TBI). Issue 7 (19th May 2020)

• Record Type: Journal Article
• Authors: Thomas, David H.; Miller, Brian; Rabinovitch, Rachel; Milgrom, Sarah; Kavanagh, Brian; Diot, Quentin; Miften, Moyed; Schubert, Leah K.
• Is Part Of: Journal of applied clinical medical physics. Volume 21:Issue 7(2020)
• Page Start: 100
• Page End: 106
• Publication Date: 2020-05-19
• Subjects: automation -- incident learning systems -- TBI; Medical physics -- Periodicals; Clinical medicine -- Periodicals; Health Physics; Clinical Medicine; Electronic journals; Periodicals; Periodicals; Fulltext; Internet Resources; 610.153
• DOI: 10.1002/acm2.12894
• Languages: English
• ISSNs: 1526-9914

Abstract

Abstract: Purpose: To identify causes of error, and present the concept of an automated technique that improves efficiency and helps to reduce transcription and manual data entry errors in the treatment planning of total body irradiation (TBI). Methods: Analysis of incidents submitted to incident learning system (ILS) was performed to identify potential avenues for improvement by implementation of automation of the manual treatment planning process for total body irradiation (TBI). Following this analysis, it became obvious that while the individual components of the TBI treatment planning process were well implemented, the manual 'bridging' of the components (transcribing data, manual data entry etc.) were leading to high potential for error. A C#‐based plug‐in treatment planning script was developed to remove the manual parts of the treatment planning workflow that were contributing to increased risk. Results: Here we present an example of the implementation of "Glue" programming, combining treatment planning C# scripts with existing spreadsheet calculation worksheets. Prior to the implementation of automation, 35 incident reports related to the TBI treatment process were submitted to the ILS over a 6‐year period, with an average of 1.4 ± 1.7 reports submitted per quarter. While no incidents reached patients, reports ranged from minor documentation issues to potential for mistreatment if not caught before delivery. Since the implementation of automated treatment planning and documentation, treatment planning time per patient, including documentation, has been reduced; from an average of 45 min pre‐automation to <20 min post‐automation. Conclusions: Manual treatment planning techniques may be well validated, but they are time‐intensive and have potential for error. Often the barrier to automating these techniques becomes the time required to "re‐code" existing solutions in unfamiliar computer languages. We present the workflow here as a proof of concept that automation may help to improve clinical efficiency and safety for special procedures.