Blind Testing Method to Validate the Efficiency of ICD-10-CM Artificial Intelligence Coding Modules.

Chih-Yen Sun; Zhenghao Li; Ming-Ju Tsai; CF Tsai; Hong-Jie Dai

doi:10.3233/shti251222

August 7, 2025

Blind Testing Method to Validate the Efficiency of ICD-10-CM Artificial Intelligence Coding Modules.

View at Publisher

Ask AI

Bookmark

Share

Key Points

MAIN FINDING: GPT-2 achieved a significant reduction in coding time compared to manual methods.
KEY EVIDENCE: The average coding time per record decreased by about 75 seconds with GPT-2 (P < 0.05).
APPROACH: The study evaluated three AI models on discharged patient data between October and November 2024.
SIGNIFICANCE: Utilizing AI in coding potentially transforms healthcare documentation, streamlining workflows.

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

This analysis compares coding time using NLP and certified coding specialists, suggesting AI can enhance efficiency.

Abstract

This study compares coding time between Natural Language Processing (NLP) and Certified Coding Specialists (CCS) using three AI models (HAN, GPT-2, and BioMistral). Analyzing discharged patients from October to November 2024, the results indicate that GPT-2 reduced the average coding time per record by approximately 75 seconds compared to manual coding (P < 0.05).

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Sun et al. (2025) studied this question.

synapsesocial.com/papers/689dfe97d61984b91e13c26c https://doi.org/10.3233/shti251222

View at Publisher

Ask AI

Bookmark

Share

Blind Testing Method to Validate the Efficiency of ICD-10-CM Artificial Intelligence Coding Modules.

Key Points

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Blind Testing Method to Validate the Efficiency of ICD-10-CM Artificial Intelligence Coding Modules.

Key Points

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading