MbShea@PWP@English@UMCP

Abstract:
The purpose of this work is to provide a cohesive and standardized procedure list for radiotherapy plan and assessment entry into the Oncospace Database, a health information system at Johns Hopkins University School of Medicine. Currently, a key issue with Oncospace is data integrity as radiotherapy plans are incorrectly annotated and pushed into the database, where they are included in training sets for a variety of machine learning projects in the broader realm of clinical informatics. This document provides clear and understandable instructions for data entry through three sections: DICOM import from Teremedica (a JHU patient records system), restoration of patient files in Pinnacle (radiation therapy planning software), and annotation and pushing of patient files. This document is aimed at summer interns between the age of 16 and 22 who will be tasked with data entry responsibilities on the Oncospace research team. With this, I hope that the interns can utilize this document on a day-to-day basis as the primary guide to radiotherapy plan entry. Longitudinally, the broader goal of this document is to improve the integrity of the data within Oncospace such that our machine learning research can be trained on more efficacious datasets.

Reader Response:
Integrating big data and artificial intelligence practices into clinical medicine represents a novel advance with the potential of improving patient care. Within the realm of radiation oncology, the Oncospace research initiative is an example of such, coordinating the actions of clinical stakeholders from across disciplines and medical centers. I maintain a level of concern on the efficacy of this process. Even if cohesive instructions are compiled to further streamline and improve the consistency of such a process, a margin of error exists. From my perspective, the effort required of coordinating data input and completing the process itself does not provide an immediate return on investment. Utilizing this document to instruct physicians on data entry additionally presents a concern as the lifetime of such a project is extremely long and it will be a while until such an effort can translate into improved patient care.

V, such an important project that will improve over time, as the newer cohorts of radiology/oncology physicians enter the profession, shape the profession, and lead on such innovation.

I do NOT know if the coding book will help here or if some norming sessions might help older participants feel (so much of this is profession pathos/ethos) more comfortable with the quality of inputs, thereby increasing some confidence.

You could make the analogy (not necessarily in your document but in discussions) to the morbidity and mortality sessions that doctors engage in....these are norming analysis sessions.