Skip to main content

JD Transcription

Dutch Cardiology Clinic Reports 30 Percent Less Note Time With Epic AI Scribe

We track medical transcription and AI scribe news. Two cardiologists at OLVG, a hospital in Amsterdam, shared their first experience with an AI scribe built into Epic. The report appeared in the November 2026 issue of the Netherlands Heart Journal. Fleur Meijer and Jonas de Jong wrote it, and both work in the cardiology department at OLVG.

What the clinic uses

An ambient AI scribe is software that listens to a visit and drafts the clinical note. It uses automatic speech recognition to turn speech into text. A large language model then writes a structured note from that text. The authors say OLVG was the first hospital in mainland Europe to adopt it at an early stage.

The workflow starts in the patient chart. A smartphone with the Epic Haiku app works as the microphone. The doctor asks for brief consent and taps the phone to start recording. When the visit ends, the audio uploads on its own. The note fills in within 15 to 60 seconds. The tool can also suggest orders. Epic deletes the audio one week after it produces the note.

About 30 percent less documentation time

The authors write that they saw about 30 percent less Epic documentation time per outpatient visit in their first evaluation. The article is a short report from the clinic. It gives the 30 percent figure as their observation and shares few study details. Clinicians also said eye contact and active listening improved right away, because the keyboard left the conversation.

The authors say chart access gives Epic an edge. A spoken word like “diuretic” can match the furosemide 20 mg once daily listed in the record. A recorder that hears the conversation alone works from the words it hears.

Slips the doctors caught

The authors saw the stomach drug pantoprazole written as “plantenzorg,” the Dutch word for plant care. They say a human scribe makes a different kind of slip. They found that drug capture varies. Sometimes it is exact, and sometimes it needs a fix.

One draft also contained the line “Advice: continue chemotherapy” after a patient described past cancer care. That advice sits outside a cardiologist’s scope. The authors also found that the history section often ran long and was hard to scan. They suggest bold text for medicines, decisions, and follow up plans.

Read every line

The authors say doctors often sign junior notes after a quick scan. They advise a fuller habit with AI drafts, because unexpected slips are easy to overlook. They write that every line needs a full read before signing, and the clinician stays fully responsible for the signed note. This review adds some time between visits, since the doctor waits for the draft and checks it.

The authors expect better notes as the model gets more chart data. They name deeper use of structured Epic records as the next step toward clear, well formatted notes.

Why it matters

This report adds a European cardiology view to the AI scribe story. It shows time savings and a strong need for careful human review. We will keep tracking medical transcription and AI scribe news as clinics share more results.

Sources