A Novel Approach to Template Filling with Automatic Speech Recognition for Healthcare Professionals

Abstract

In the evolving landscape of medical documentation, the necessity for efficient and accurate record-keeping systems is paramount, especially in specialised fields such as neurology where precision in terminology is crucial. This paper introduces a pioneering application of a fine-tuned Whisper model, specifically adapted for brain-related medical terms, integrated with an AI-driven system for automated template filling. The proposed system leverages advanced speech recognition technologies to capture doctors' verbal inputs and accurately transcribe these into designated report templates. The process simplifies the documentation workflow, significantly reducing the cognitive and administrative load on healthcare providers by enabling them to focus more on patient care rather than paperwork. Our research details the development and implementation of this innovative system, including the specific adaptations made to the Whisper model to enhance its accuracy with neurology-specific terminology. We also evaluate the system's performance in real-world medical settings and discuss the practical implications of integrating such AI tools in clinical practice. Furthermore, the system's capacity to generate ready-to- print PDF reports not only streamlines the documentation process but also ensures consistency and reliability in medical records. The overarching aim of this project is to demonstrate how targeted AI solutions can address the unique challenges of medical documentation, offering substantial benefits to healthcare providers and patients alike.