A Systematic Review of Artificial Intelligence Applications in Healthcare Documentation: Challenges and Impact on Care Quality (Preprint)

BACKGROUND Healthcare documentation is a critical component of patient care, yet it frequently suffers from incompleteness, inaccuracy, and inefficiency. Artificial intelligence (AI) has emerged as a promising tool to improve documentation quality and reduce workload. OBJECTIVE This systematic review aimed to evaluate the applications of AI in healthcare documentation, identify associated challenges, and assess its impact on care quality. METHODS A systematic search was conducted in the Web of Science Core Collection, yielding 2,862 records. After removing 1,752 duplicates, 1,110 records were screened. Following title and abstract screening, 116 studies met the inclusion criteria. Due to methodological heterogeneity, a narrative synthesis approach was used. Quality assessment was performed using a 9-point scale covering six key domains. RESULTS AI applications in healthcare documentation were categorized into six domains: time saving and workload reduction, accuracy and safety, role differences between physicians and nurses, user experience and patient interaction, required infrastructures, and technical/localization challenges. Ambient voice technology and large language models (LLMs) reduced documentation time by 27–80.8%. However, issues such as model hallucination, privacy concerns, and lack of linguistic localization were prevalent. The most effective approaches combined automatic speech recognition (ASR), LLMs, retrieval-augmented generation (RAG), local deployment, and expert review. CONCLUSIONS AI holds significant potential to improve healthcare documentation efficiency and quality, but successful implementation requires addressing technical, ethical, linguistic, and organizational challenges. A hybrid approach combining AI with human review and validation is recommended.