Improving documentation quality and patient interaction with AI: a tool for transforming medical records—an experience report

Abstract

Background: Physician burnout is closely linked to the excessive time spent on documentation tasks, which often compromises the quality of medical records and patient care. Voa, an AI-driven tool tailored to the specific needs of Brazilian healthcare, converts audio from medical consultations into structured clinical documents, reducing documentation time and improving record quality. By integrating technologies such as Whisper, a speech recognition system, and generative AI for text structuring, Voa seeks to minimize ‘copy and paste’ errors, reduce the time spent on medical history generation, and enhance doctor-patient interactions. Methods: From March to August 5, 2024, the effectiveness of Voa was evaluated using quantitative and qualitative metrics. Quantitative measures included the number of documents generated, total users, and the weekly activation rate, which measures the percentage of new users generating at least one document within a week of registration. Qualitative metrics were collected through user satisfaction surveys, including the Net Promoter Score (NPS) and Customer Satisfaction Score (CSAT), along with feedback analysis. Results: By August 2024, Voa had generated 24,654 documents and registered 2,006 users, with daily document generation peaking at 504. The weekly activation rate stabilized above 40% by April. User satisfaction improved steadily, with NPS rising from 18 in April to 58 in July. A total of 84% of users rated their experience as 4 or 5 out of 5 on the CSAT scale, with feedback highlighting the platform’s positive impact on clinical workflows and suggestions for further refinements. Conclusions: Voa has been validated as an effective generative AI tool, offered as a software as a service (SaaS) solution, for improving medical record creation and reducing the documentation burden on physicians. Its growing user base, coupled with high satisfaction metrics, supports its scalability and potential to improve healthcare delivery by allowing doctors to focus more on patient care. The results on Activation Rate reflected consistent engagement from new users and the results on NPS and CSAT indicates strong user advocacy. Future refinements, including enhanced transcription accuracy and customization options, will further optimize the tool’s performance. Additional studies, such as randomized controlled trials, are important to solidify the tool’s effectiveness in diverse healthcare environments and ensure its adaptation to linguistic and medical terminology nuances.

Ähnliche Arbeiten