Automating the overburdened clinical coding system: challenges and next steps

Venkatesh, Kaushik P.; Raza, Marium M.; Kvedar, Joseph C.

doi:10.1038/s41746-023-00768-0

Cited by 16 publications

(7 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This variability can arise, for example, from particular selection of wording in notes by clinicians, or code assignment biases by providers or medical coders. Opportunities for improving ICD code use can occur through improved education of providers and coders but also through use of natural language processing or other algorithm‐based strategies 22 . Other terminologies, for example, SNOMED‐CT, are more conducive to research purposes and have a better structure, more comprehensive breadth, and more interoperability between different terminologies.…”

Section: Discussionmentioning

confidence: 99%

“…Opportunities for improving ICD code use can occur through improved education of providers and coders but also through use of natural language processing or other algorithmbased strategies. 22 Other terminologies, for example, SNOMED-CT, are more conducive to research purposes and have a better structure, more comprehensive breadth, and more interoperability between different terminologies. However, the ubiquity of ICD-10-CM code use in the United States is unmatched by any other terminology.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

An introduction to ICD code development for pediatric neurology

Baker

Bonkowsky

2023

Annals of the Child Neurology Society

View full text Add to dashboard Cite

ObjectiveQuerying large data sets in the United States is challenging due to limitations of International Classification of Diseases 10th edition Clinical Modification (ICD‐10‐CM) codes. ICD codes were developed for tracking mortality and for billing purposes but are also the most widely used data structure to represent clinically significant and distinct disorders. We report an approach for developing new ICD codes based on our work creating new codes for leukodystrophies.MethodsImportant steps in ICD‐10‐CM code development include working with the ICD‐10‐CM Coordination and Maintenance Committee, a subset of the National Center of Health Statistics (NCHS); working to ensure the code has the best placement and is appropriate for submission; presenting the code and accompanying proposal (rationale) at a Coordination and Maintenance Committee Meeting; and requesting letters of support and addressing concerns raised by various groups.ResultsWe describe the historical development of ICD codes as well as their current hierarchical structure. Important features for successful code development included consulting future ICD‐11‐CM code structure; determining which codes are most important to the community; having a multidisciplinary approach; and obtaining organizational and institutional support.ConclusionFocusing on the clinical importance of leukodystrophy codes was important for their approval. Although challenging, there is a route for new code development, and we were ultimately successful in obtaining approval for 10 new leukodystrophy ICD‐10‐CM codes. Understanding ICD codes and their structure, considering their usage for clinical and research work, and appreciating how new codes can be developed is important for the pediatric neurology community.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

An introduction to ICD code development for pediatric neurology

Baker

Bonkowsky

2023

Annals of the Child Neurology Society

View full text Add to dashboard Cite

show abstract

“…Automated Medical Coding (AMC) is the idea that artificial intelligence can automate clinical coding. In recent years, there has been a significant increase in AMCrelated work [14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32] through deep learning. Although research in this field has grown, this problem is far from being solved [18,33].For instance, automated coding remains a complex problem because extracting knowledge from patients' clinical records is challenging.…”

Section: Background and Significancementioning

confidence: 99%

Medical Diagnosis Coding Automation: Similarity Search vs. Generative AI

Klotzman

2024

Preprint

View full text Add to dashboard Cite

Objective: This study aims to predict ICD-10-CM codes for medical diagnoses from short diagnosis descriptions and compare two distinct approaches: similarity search and using a generative model with few-shot learning. Materials and Methods: The text-embedding-ada-002 model was used to embed textual descriptions of 2023 ICD-10-CM diagnosis codes, provided by the Cen-ters provided for Medicare & Medicaid Services. GPT-4 used few-shot learning. Both models underwent performance testing on 666 data points from the eICU Collaborative Research Database. Results: The text-embedding-ada-002 model successfully identified the relevant code from a set of similar codes 80 percent of the time, while GPT-4 achieved a 50 percent accuracy in predicting the correct code. Discussion: The work implies that text-embedding-ada-002 could automate medical coding better than GPT-4, highlighting potential limitations of generative language models for complicated tasks like this. Conclusion: The research shows that text-embedding-ada-002 outperforms GPT- 4 in medical coding, highlighting embedding models usefulness in the domain of medical coding.

show abstract

“…The increasing workload due to the growing number of patient records makes medical coding an intensive task [3]. Advanced technologies such as artificial intelligence (AI) and natural language processing (NLP) can automate or semi-automate coding, reducing the burden on human coders and improving efficiency [7,8]. For example, AI-powered systems can analyse clinical records such as doctor's notes and medical records, as well as historical coding data, to suggest appropriate codes based on the documented information [8].…”

Section: Introductionmentioning

confidence: 99%

“…Advanced technologies such as artificial intelligence (AI) and natural language processing (NLP) can automate or semi-automate coding, reducing the burden on human coders and improving efficiency [7,8]. For example, AI-powered systems can analyse clinical records such as doctor's notes and medical records, as well as historical coding data, to suggest appropriate codes based on the documented information [8]. They can also compare coded information with clinical documentation to identify discrepancies or potential errors, ensuring coding integrity and compliance with coding guidelines.…”

Section: Introductionmentioning

confidence: 99%

Leveraging Language Models for Inpatient Diagnosis Coding

Suvirat,

Tanasanchonnakul,

Chairat

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

Medical coding plays an essential role in medical billing, health resource planning, clinical research and quality assessment. Automated coding systems offer promising solutions to streamline the coding process, improve accuracy and reduce the burden on medical coders. To date, there has been limited research focusing on inpatient diagnosis coding using an extensive comprehensive dataset and encompassing the full ICD-10 code sets. In this study, we investigate the use of language models for coding inpatient diagnoses and examine their performance using an institutional dataset comprising 230,645 inpatient admissions and 8677 diagnosis codes spanning over a six-year period. A total of three language models, including two general-purpose models and a domain-specific model, were evaluated and compared. The results show competitive performance among the models, with the domain-specific model achieving the highest micro-averaged F1 score of 0.7821 and the highest mean average precision of 0.8097. Model performance varied by disease and condition, with diagnosis codes with larger sample sizes producing better results. The rarity of certain diseases and conditions posed challenges to accurate coding. The results also indicated the potential difficulties of the model with long clinical documents. Our models demonstrated the ability to capture relevant associations between diagnoses. This study advances the understanding of language models for inpatient diagnosis coding and provides insights into the extent to which the models can be used.

show abstract

Automating the overburdened clinical coding system: challenges and next steps

Cited by 16 publications

References 16 publications

An introduction to ICD code development for pediatric neurology

An introduction to ICD code development for pediatric neurology

Medical Diagnosis Coding Automation: Similarity Search vs. Generative AI

Leveraging Language Models for Inpatient Diagnosis Coding

Contact Info

Product

Resources

About