MIRA-Medical Information Review Assistant

To allow physicians to devote more energy to patient care, our hospital's Artificial Intelligence Center, together with a clinical physician advisory team, has successfully developed the "MIRA" (Medical Information Review Assistant). This system integrates the latest automatic speech recognition (ASR) and large language model (LLM) technologies to achieve several breakthrough functions:

1. Real-time conversion of doctor-patient dialogues into structured medical records
2. Automatic checking and correction of grammatical errors
3. Assistance with inferring diagnostic key points
4. Suggestions for supplementary medical information

The system's benefits have been impressively verified in clinical practice: in the short three-month period from May to August 2024, it accumulated over 1,500 actual application cases. Most amazingly, the system successfully reduced traditional medical record writing that typically requires 20-30 minutes to just 5-6 minutes, improving efficiency by up to 80%! This not only saves physicians valuable time but also allows them to focus more on patient diagnosis and treatment without being distracted by medical record writing, truly enhancing overall healthcare service quality.

Nursing Voice Documentation

This product is specifically designed and tailored for clinical nursing environments. Equipped with nursing-specific speech recognition technology and integrating advanced AI language models, it converts spoken voice input from nursing staff into structured electronic medical records. This streamlines tedious documentation time and enables clinical nurses to complete records in real time.

It not only reduces documentation time for clinical nurses but also enhances the completeness of clinical records. This allows nursing staff to focus more intently on patient care, effectively alleviating their clinical workload burden.

This system has demonstrated exceptional technical performance in practical applications: it achieves a word error rate (WER) of only 14.47%, and processes a 13-minute audio file in just 63 seconds. Moreover, its intelligent auto-correction feature ensures high-accuracy recognition of specialized medical terminology.

In clinical settings, the system significantly optimizes workflow. Previously, nurses required an additional hour post-handover to complete nursing records. With the implementation of this technology, standardized nursing documentation (DART) can be completed within 15 minutes, substantially enhancing the completeness, consistency, and accuracy of nursing records.

Journal Article

• HCode-switching automatic speech recognition for nursing record documentation: system development and evaluation.
  https://nursing.jmir.org/2022/1/e37562/

Awards

2023

• National Innovation Award - Clinical Innovation Award

Patents

• Taiwan Patent Certificate ( No.M663031)

CTCAE：Based Drug Side Effects Assessment and Education System with AI Assistance

To enhance cancer patient care quality, our hospital's Artificial Intelligence Center collaborated innovatively with the cancer ward medical team to successfully develop the "Chemotherapy Assistant (CTCAE)," utilizing the Google Vertex AI platform and large language model (LLM) technology to help healthcare professionals conduct more precise chemotherapy side effect assessments and patient education.

The system not only automatically generates side effect tracking lists based on patients' chemotherapy prescriptions, allowing healthcare providers to perform personalized tracking and thoroughly document various side effects and adverse reactions, but also serves as a resource for medication information inquiries and educational guidance. Through real-time, precise information support, it significantly reduces the administrative burden on the medical team, enabling them to effectively focus on patient care and further enhance overall cancer treatment outcomes.

Since its launch, the CTCAE system has continuously collected valuable feedback from hospital staff for optimization, including adding new features and expanding the chemotherapy medication information database to better align with clinical practice needs. According to user surveys, the system's effectiveness has exceeded expectations: successfully reducing the original 20-30 minutes required for interviews and documentation to just 5-10 minutes, saving up to 85% of time! This breakthrough achievement not only allows healthcare professionals to quickly focus on tracking priorities and reduce paperwork burden, but also enables nurses to invest valuable time in patient care, truly realizing the concepts of smart healthcare and patient-centered care.

Contact window 

Project Management Group
Phone:+886-422052121 ext 12534