Technical note: An automated document verification tool in radiation oncology EMR: Application for LDR prostate brachytherapy

Junliang Xu, Baoshe Zhang, Mariana Guerrero, Chaitanya Kalavagunta, Shifeng Chen, Huijun XuJournal of Applied Clinical Medical Physics 2024 · DOI

TL;DR

This paper introduces a novel, script-based automated tool to verify documentation for Low-Dose-Rate (LDR) prostate brachytherapy. By leveraging Python scripts integrated with the Electronic Medical Record (EMR) system, the tool significantly improves efficiency and accuracy in routine chart checks and audits, reducing manual effort and potential human errors.

What Problem Does This Paper Address?

Radiation oncology, particularly LDR prostate brachytherapy, relies heavily on meticulous documentation for patient safety and treatment efficacy. Traditionally, this involves extensive manual chart checks by physicists and other medical professionals. This process is time-consuming, prone to human error (typos, inconsistencies), and can contribute to burnout. While automation has been explored for other radiation therapy modalities like External Beam Radiotherapy (EBRT) and High-Dose-Rate (HDR) brachytherapy, its application to LDR prostate brachytherapy documentation verification remained largely unexplored. This paper aims to fill that gap by developing and evaluating an automated tool specifically for this purpose.

Methods

The authors developed an in-house Python script-based tool designed to automatically verify specific checklists within the Electronic Medical Record (EMR) system, aligning with their institutional practice guidelines for prostate seed implants (PSI). The tool is compatible with Word and PDF document formats and can operate in two modes: standalone (manual initiation) and integration (automated workflow).

The integration mode begins by identifying patients with active Care Paths in the Varian Aria EMR system. It then extracts and parses specific EMR documents from the Radiation Oncology Information System (RO-IS) database. The verification process employs a multi-layered approach:

  1. Data Completeness Check: Ensures all mandatory fields are filled and meet expectations (e.g., verifying "source activity").
  2. Consistency Check: Compares data points across different documents to identify discrepancies (e.g., implanted seed numbers in quality checklists vs. survey reports).
  3. Allowable Value Range Check: Confirms data points fall within established institutional protocols or clinical guidelines (e.g., seed quantities against a nomogram with a tolerance limit).

The scripts were validated using test cases derived from previous patient data, intentionally introducing errors to check for false negatives and ensuring correct flagging for false positives. A limitation noted is the inability of the current scripts to interpret images or screenshots within documents.

Datasets

The study evaluated the automated tool on a random cohort of 50 LDR prostate brachytherapy patients. The tool was initially optimized based on data from the previous 400 patients within their institution. The paper also references institutional nomograms (Table 2) used as a guideline for PSI seed ordering, which the tool utilizes for validation. No new benchmarks were introduced; rather, the tool was applied to existing patient EMR data.

Key Insights and Ideas

  • Bridging the Automation Gap: This work successfully extends the benefits of automation in documentation verification from EBRT and HDR brachytherapy to the less-explored domain of LDR prostate brachytherapy.
  • Efficiency Beyond Speed: While time savings are significant, the primary benefit lies in enhanced data accuracy and integrity, reducing the likelihood of human errors that could impact patient care.
  • Targeted Automation for Specific Workflows: The tool demonstrates particular effectiveness in streamlining critical, high-volume tasks like seed ordering and monthly audits, which are prone to manual errors.
  • Balancing Automation and Clinical Judgment: The authors acknowledge that automation cannot entirely replace human oversight. They emphasize the importance of physician discretion, especially in complex decision-making processes like seed number calculation, where patient-specific factors are crucial.
  • Foundation for Broader Applications: The script-based approach is adaptable and can potentially be extended to other EMR systems and different types of documentation checks within radiation oncology.

Key Results

The automated tool was applied to a cohort of 50 LDR patients, demonstrating significant improvements:

  • Error Detection:
    • Seed Ordering: Identified typographical errors, information inconsistencies, and incompleteness in 7 out of 50 patients.
  • Monthly Audit: Flagged two errors, including incorrect survey meter data and omissions in quality checklists.
  • Time Savings:
    • Seed Ordering: Reduced manual chart-checking time by an average of 5 minutes per patient.
  • Monthly Audit: Achieved an average time saving of 10 minutes per audit.
  • Accuracy: The tool substantially improved the accuracy and speed of document validation compared to conventional manual checks.

Limitations and Future Work

  • Image Interpretation: The current version of the tool cannot process or interpret images, such as treatment reports embedded as PNG files in EMR documents. This limits its ability to check checklist items that rely on visual information.
  • FMEA Application: The authors note the lack of a comprehensive Failure Mode and Effects Analysis (FMEA) tailored for LDR brachytherapy documentation verification, citing limited relevant literature.
  • Future Directions:
    • Develop more robust scripting capabilities, potentially including image analysis.
  • Conduct a comprehensive FMEA specifically for LDR brachytherapy documentation requirements.
  • Explore integration with other EMR systems and broaden the scope of automated checks.
  • Continue to refine the tool for user-friendliness and to address gaps across different clinical scenarios.

Summary updated

Abstract

The implementation of an automated tool tailored for LDR prostate brachytherapy has demonstrated its efficiency benefits. Such an approach can help other clinics substantially enhance routine chart checks, periodic audits, and other applications in similar clinical settings.
BibTeX
@article{xu2024technical,
  author = {Junliang Xu and Baoshe Zhang and Mariana Guerrero and Chaitanya Kalavagunta and Shifeng Chen and Huijun Xu},
  title = {{Technical note: An automated document verification tool in radiation oncology EMR: Application for LDR prostate brachytherapy}},
  journal = {Journal of Applied Clinical Medical Physics},
  year = {2024},
  doi = {10.1002/acm2.14466},
}
Figure