Medical Coding Experts’ Perspectives on the Challenges, Benefits, and Acceptance of Artificial Intelligence Technologies in ICD Coding: A Sequential Multi-Phase Study Protocol

Background The International Classification of Diseases (ICD) is an essential component of health information management systems and provides a standardized approach for representing diseases, health conditions, and clinical events. ICD coding contributes to multiple healthcare processes, including clinical documentation management, reimbursement, epidemiological surveillance, healthcare planning, quality assessment, and clinical research. Despite its importance, traditional manual coding processes face increasing challenges due to the rapid expansion of electronic health records (EHRs), growing complexity of clinical documentation, and increasing demand for timely and accurate coded data. These challenges may result in increased workload among coding professionals, longer processing times, variations in coding decisions, and potential inconsistencies in coded information. Recent developments in artificial intelligence (AI), including machine learning (ML), clinical natural language processing (NLP), and large language models (LLMs), have accelerated the development of automated clinical coding (ACC) and computer-assisted coding (CAC) systems. These technologies have demonstrated the ability to extract clinical concepts from unstructured medical texts, support code assignment, and improve the efficiency of coding workflows. However, successful integration of AI-based coding systems requires consideration of factors beyond technical performance, including users’ perceptions, trust, workflow compatibility, organizational readiness, and professional acceptance. Medical coding experts represent the primary professional group involved in ICD coding activities and will directly interact with AI-supported coding technologies. Their practical experience with documentation challenges, coding workflows, and existing organizational processes provides valuable insight into factors that may facilitate or hinder AI implementation. Although previous studies have extensively examined algorithm development and predictive performance of AI coding models, limited attention has been given to the perspectives of end users who will adopt these technologies in real healthcare environments. In Iran, despite increasing interest in artificial intelligence applications within healthcare, evidence regarding the perceptions, concerns, and acceptance of AI-assisted coding technologies among health information management professionals remains limited. Understanding these perspectives before implementation may help ensure that AI solutions are developed and introduced in accordance with professional requirements and healthcare organizational conditions. Therefore, this study aims to investigate medical coding experts’ perspectives regarding the perceived benefits, implementation challenges, and acceptance of artificial intelligence technologies in ICD coding processes in teaching hospitals affiliated with Mashhad University of Medical Sciences (MUMS), Iran. Methods This sequential multi-phase study will be conducted during ۲۰۲۵–۲۰۲۶ in teaching hospitals affiliated with Mashhad University of Medical Sciences, Iran. The study will include three main phases: (۱) literature review and conceptual framework development, (۲) questionnaire development and psychometric evaluation, and (۳) cross-sectional assessment of medical coding experts’ perspectives. In the first phase, a comprehensive literature review will be performed to identify artificial intelligence applications in ICD coding, including automated clinical coding and computer-assisted coding approaches, reported benefits, implementation barriers, and factors influencing technology acceptance. Searches will be conducted in PubMed, Scopus, Web of Science, and Google Scholar databases. Findings from this phase will guide the development of the conceptual framework and questionnaire items. In the second phase, a structured questionnaire will be developed based on evidence obtained from the literature review and established technology acceptance frameworks, including the Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology (UTAUT), and Technology Readiness Index (TRI). The developed instrument will assess perceived benefits, perceived challenges, and acceptance of AI-assisted ICD coding technologies. Content validity and reliability of the instrument will be evaluated before implementation in the main survey. In the third phase, a census survey will be conducted among eligible medical coding experts working in health information management departments of teaching hospitals affiliated with MUMS. Data will be collected using electronic and paper-based questionnaires. Descriptive and inferential statistical analyses will be performed to evaluate perceived benefits, challenges, acceptance levels, and factors associated with intention to use AI-assisted coding technologies. Expected Results This study is expected to provide evidence regarding medical coding experts’ perspectives toward artificial intelligence technologies in ICD coding processes. The findings may identify perceived advantages of AI-assisted coding systems, including potential improvements in coding accuracy, reduction of workload, increased efficiency, and enhancement of healthcare data quality. Furthermore, the study will explore barriers that may influence AI adoption, including concerns related to algorithm transparency, reliability of AI-generated recommendations, professional role changes, technical infrastructure, clinical data quality, training requirements, and organizational preparedness. Conclusion The implementation of artificial intelligence in ICD coding requires consideration of both technological capabilities and the perspectives of professionals involved in coding workflows. This study will generate evidence regarding the factors affecting acceptance of AI-assisted coding technologies among medical coding experts in teaching hospitals affiliated with Mashhad University of Medical Sciences. The findings may support health information managers, hospital administrators, and policymakers in developing appropriate strategies for AI implementation, workforce preparation, and user-centered integration of artificial intelligence within health information management systems.

کلیدواژه ها:

Artificial Intelligence ، Automated Clinical Coding ، Computer-Assisted Coding ، ICD Coding ، Health Information Management ، Medical Coding Experts ، Technology Acceptance Model ، Teaching Hospitals