NUH's AI Scribes Give Singapore Nurses Back 16 Minutes Per Patient Encounter

Singapore's nurses spend a large share of every shift reconstructing the work they just did — typing up notes, chasing handovers, rebuilding documentation long after a patient encounter ends. As the country's healthcare system absorbs heavier demand, that paperwork has quietly become one of the profession's biggest drains on time and morale. National University Hospital (NUH) is now treating the problem as an engineering challenge, moving AI transcription from pilot into everyday nursing workflows.

Speaking at HealthTechX Asia 2026, NUH chief nurse Karen Koh detailed how the hospital runs two complementary speech-to-text tools: MediVoice, an in-house AI transcription system for clinical documentation, and Scribe, a transcription tool from Open Government Products for non-clinical notes. Both handle multiple languages and dialects, and generate structured documentation through customised templates under institutional safeguards. Early data from selected use cases shows time savings of up to 16 minutes per patient encounter — time that accumulates across shifts into more direct patient care, mentoring of junior colleagues, and less unnecessary overtime.

The transcription push is one layer of a broader nursing-transformation stack NUH has assembled over the past few years: electronic medical records for handovers, mobile apps for patient identification, automated vital-signs capture, smart infusion pumps, medication-management systems and robotic process automation for inventory checks. The common thread is that technology is embedded into existing workflows rather than bolted on as a separate process — an approach Koh argues is essential for adoption in a profession where burnout is as much about whether the work feels meaningful as about hours worked.

Equally notable is how NUH has managed the human side of deployment. Nurse champions act as cultural ambassadors across departments, and frontline nurses co-develop the tools with engineers rather than having them imposed from above. Governance is treated as a precondition, not a compliance afterthought: risks are evaluated before deployment, safeguards are built in throughout implementation, and human review stays central to clinical documentation. "The key is that we automate the routine and elevate the human," Koh said.

Why it matters for Singapore: Nursing documentation burden is a retention issue as much as an efficiency one, and Singapore's public healthcare system faces rising demand alongside a tight talent pipeline. NUH's pairing of an in-house clinical tool with an OGP-built government product also shows how Singapore's govtech ecosystem can flow into clinical settings. The hospital is now exploring ambient AI, intelligent patient interfaces and AI-assisted scheduling — early signs that Singapore's hospitals are moving from pilots to governed, scaled deployment of frontline AI.