NLP Correction Clarifies Medical Coding Complexity Figures
A correction published in JMIR Med Inform clarifies misprinted y-axis labels in figures from the 2023 paper "An End-to-End Natural Language Processing Application for Prediction of Medical Case Coding Complexity" by Xu et al. The notice states two figure label errors affected the presentation of complexity rating comparisons, specifically in the originally published Figure 9. The underlying data, model outputs, and the paper's conclusions remain unchanged. The correction supplies a revised figure and affirms the validation results and expert-comparison analyses are intact. This is a presentation-level erratum, not a change to algorithms, data, or validation metrics.
What happened
A corrigendum in JMIR Med Inform corrects misprinted y-axis labels in the 2023 paper "An End-to-End Natural Language Processing Application for Prediction of Medical Case Coding Complexity" by Xu et al. The authors issued two corrections that affect how complexity rating comparisons were labeled in the originally published Figure 9. The correction replaces the misprinted figure with a revised version and explicitly states that the underlying data and the paper's conclusions remain unchanged.
Technical details
The error is a visualization labeling issue: y-axis labels in two subpanels did not correspond to the intended numeric values submitted with the manuscript. The affected panels compared expert coder ratings and the validation model's predictions on a gold-standard set, both for raw four-level complexity scores and a binary grouping into simple versus complex cases. No changes were made to the dataset, model code, evaluation pipeline, or numerical results reported in tables and text. Practitioners relying on the paper's metrics and performance numbers can continue to use the published figures after the label correction.
Context and significance
This paper presents an applied natural language processing pipeline for predicting coding complexity, including human expert agreement analyses and a validation model comparison. Visualization fidelity matters for reproducibility and interpretability, especially when readers inspect calibration, inter-rater agreement, or thresholding behavior visually. The correction preserves scientific integrity by ensuring that readers and downstream implementers interpret plots correctly, but it does not alter the model, training regimen, or validation outcomes.
What to watch
Confirm the revised figure when citing visual comparisons from this study. If you reproduced any plots or used the original images for downstream analysis, re-check your interpretations against the corrected figure to avoid propagation of the labeling error.
Scoring Rationale
This is a presentation-level correction that does not change data, methods, or conclusions. It matters for clarity and reproducibility but has limited technical impact on practitioners or the field.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problemsStep-by-step roadmaps from zero to job-ready — curated courses, salary data, and the exact learning order that gets you hired.
