SingaporeHealthcare and life sciences2024 – ongoingNational University Health System
Russell-GPT information extraction for medical coders
NUHS put Russell-GPT into use for about 6 to 8% of medical-coding workload
Problem
Medical coders read clinical records before assigning billing and subsidy codes. Extracting relevant information slowed the work.
Approach
NUHS built Russell-GPT to extract and summarise information for human coders. It piloted the system in October 2024 and went live in August 2025 on the simplest cases. Processing stayed on NUHS servers.
Results
about 6 to 8 per centshare of total workload covered by Russell-GPT
about 90 per centaccuracy reported by NUHS; human oversight remained necessary
Takeaway
Coders still assigned the codes; the 25% time-saving goal remained a future target.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Internal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
[1]National University Health System9 Mar 2026
NUHS reported limited live deployment and about 90% accuracy. The system covered about 6 to 8% of workload; 25% time savings remained a goal for wider rollout.
Full details are in the original sources.
Added 1 Nov 2026 · Updated 1 Nov 2026
