97.1% Medical Terminology Recognition Accuracy… GC MediAI's 'Doctor’s Companion AI'

"Internal Data Processing Infrastructure, an Advantage for Security"

"The core of artificial intelligence (AI) that transcribes conversations in the examination room to electronic charts ultimately depends on how accurately it understands what the doctor says. The proprietary speech recognition model applied to 'Uisarang AI' boasts a medical terminology recognition accuracy rate of 97.1%. This figure surpasses the mid-90% range of general-purpose cloud speech-to-text (STT) models and even exceeds the 90% range of open-source models."

At the 'GC MediAI Clinical Solution Uisarang AI Media Day' held on the 26th at the Conrad Hotel in Yeongdeungpo-gu, Seoul, Kim Jintae, CEO of GC MediAI, highlighted the strengths of GC MediAI's latest product Uisarang AI by emphasizing its "outstanding performance in recognizing medical terms."

GC MediAI's flagship product, the electronic medical record (EMR) solution 'Uisarang,' is the number one electronic chart used by more than 16,000 medical institutions nationwide. Earlier this year, GC MediAI launched Uisarang AI, a new product that integrates AI capabilities. Uisarang AI is an AI clinical solution that instantly transcribes examination conversations in real time and pushes them directly into the EMR chart. It transcribes conversations between healthcare professionals and patients using speech recognition, then structures the information in the SOAP (Subjective, Objective, Assessment, Plan) format, creating a seamless workflow that connects pre-exam patient summaries, post-exam follow-up management, and the viewing of clinic management indicators.

CEO Kim stated, "Uisarang AI is a product that combines decades of electronic chart operating know-how with medical AI technology, transforming the way work is done in examination rooms," adding, "By reducing the time spent on chart writing and code searches, we aim to set a new standard for the 'Next Appointment Room,' enabling healthcare professionals to focus on their core task—caring for and engaging with patients."

Uisarang AI consists of five main features: ▲ "STT·SOAP Charting," which transcribes examination conversations in real time and organizes them in the SOAP format specialized for medical terms ▲ "Past Consultation Summary," which extracts only the key information from accumulated medical records and test results ▲ "AI Diagnosis·Prescription Search," which analyzes consultation content to suggest appropriate diagnosis codes and prescription candidates ▲ "AI Assistant," which responds based on patient history ▲ "Uisarang Insight," which visualizes operational metrics such as trends in patient numbers. The AI helps not only with chart writing, but also with selecting billing codes.

To address the data security concerns that independent clinics are most sensitive about, CEO Kim highlighted the use of an in-house infrastructure. All medical data, including examination conversations and consultation records, is processed only within the internal infrastructure and is never sent out to external AI services. As all processes from speech recognition to summarization and structuring are handled by proprietary models, rather than by calling the API of a general-purpose large language model (LLM), this structure is advantageous for security.

CEO Kim said, "With the adoption of Uisarang AI, we plan to expand EMR from a 'system for documentation and billing' into a foundation that supports the entirety of clinical practice and clinic management." He envisions developing into a "medical OS" that connects to external services via cloud EMR by March 2027.

This content was produced with the assistance of AI translation services.