Using Generative Models to Improve Clinical Documentation Accuracy

EasyChair Preprint no. 14155

Abstract

Clinical documentation is critical for patient care, billing, and research, yet its accuracy is often compromised due to variability in language, incomplete data entry, and clinician workload. This paper explores the application of generative models, particularly advanced machine learning techniques, to enhance the accuracy of clinical documentation. We propose a framework that leverages generative models, such as transformers and generative adversarial networks (GANs), to automate and refine the documentation process. Our approach includes training these models on diverse and extensive healthcare datasets to capture medical terminologies, nuances, and contextual information. We demonstrate that generative models can generate coherent and contextually relevant documentation, reduce errors, and ensure compliance with clinical standards. Additionally, we discuss integration strategies with existing electronic health record (EHR) systems and evaluate the impact on clinical workflow efficiency and data quality. The findings indicate significant improvements in documentation accuracy, which could lead to better patient outcomes, reduced administrative burden, and enhanced overall healthcare delivery. This study underscores the potential of generative models as transformative tools in the domain of clinical documentation.

Keyphrases: Clinical Documentation, Clinical Workflows, Data Security, generative models, medical coding, Patient privacy

@Booklet{EasyChair:14155,
  author = {John Owen},
  title = {Using Generative Models to Improve Clinical Documentation Accuracy},
  howpublished = {EasyChair Preprint no. 14155},

  year = {EasyChair, 2024}}