-
Technology
-

AI Pioneer Lands Major Smart Hospital Deal to Streamline Clinical Documentation

By
Distilled Post Editorial Team

The photograph released after Thursday’s signing in Seoul is a familiar corporate ritual. Executives from a hospital and a technology supplier stand shoulder to shoulder, holding up a signed document for the cameras. Yet the memorandum of understanding between Samsung Medical Center and Heidi, the Melbourne-founded ambient documentation company, contains a more interesting proposition than the usual announcement of a commercial partnership. Neither party is making a substantial binding commitment. Heidi will test its transcription software on the hospital’s own servers, connecting it to DARWIN, the electronic medical record platform. The two organisations will assess its technical performance and whether doctors actually choose to use it. Samsung Medical Center will become Heidi’s first Korean partner and, if the pilot succeeds, a reference site for other hospitals considering the technology.

The significance lies in what the organisations intend to establish beyond the pilot. They plan to develop a Korean Clinical Documentation Standard, accompanied by an implementation guide for the FHIR data-exchange format. A reference architecture for future deployments is also on the agenda. In effect, one hospital is using an initial technology trial to help determine what a clinical note should look like, how it should be structured and how it should enter the medical record before the supplier moves into a wider Korean market. The commercial logic is straightforward. Heidi’s chief executive has pointed to the extraordinary patient volumes handled by Korean clinicians, arguing that time spent typing records is time diverted from the patient sitting in front of them.

For Britain, the arrangement raises a question about who gets to shape the infrastructure behind a rapidly expanding clinical technology market. Ambient voice technology has travelled quickly from experimental pilots to an expected feature of NHS modernisation in England. NHS England published its self-certified supplier registry in January, listing 19 companies. By October, the number had risen to around 30, including Heidi. Updated national guidance and clarification from the medicines regulator on the application of medical device legislation to these products followed on 29 July. The wider national framework encourages NHS organisations to adopt the technology at pace, within a broader £10bn technology investment.

Yet the supplier registry is not a procurement framework, and NHS England makes clear that inclusion does not amount to an endorsement. Individual NHS organisations remain responsible for purchasing decisions under their own governance arrangements. That decentralised approach has already generated substantial contracts, including the £1.9m award made by two East Midlands acute trusts to Accurx. But it leaves many of the questions being addressed in Seoul to be negotiated separately by hospitals, trusts and practices across England.

The consequences are significant. What information should a structured clinical note contain? How should an ambient tool transfer its output into an electronic patient record? Where should recordings and transcripts be stored, and who controls the information generated by the software? These are fundamental questions about clinical practice and information governance. In England, they are largely being settled through individual purchasing decisions, with no guarantee that different organisations will arrive at compatible answers.

The imbalance in negotiating power is difficult to ignore. A supplier dealing with dozens of separate buyers can develop its own documentation templates, integration methods and contractual terms, then present substantially the same arrangements to each new customer. Each trust, meanwhile, must negotiate within its own financial and technical constraints. Samsung Medical Center enters the relationship from a stronger position. It controls its electronic record environment and carries the institutional standing to make itself an attractive reference site. That gives it grounds to demand local hosting and a role in setting national documentation standards. Few NHS organisations can command comparable influence. The fragmented electronic record market makes common standards harder to establish, while the absence of a central negotiating position allows suppliers to shape practice through repeated individual agreements.

There is, of course, no guarantee that Seoul’s approach will succeed. The memorandum is non-binding, the pilot has yet to produce results and the stated ambitions come from the organisations involved in signing it. The wider evidence supporting ambient AI adoption also deserves careful examination. A survey published by Heidi in July found that 88 per cent of clinicians identified documentation as their greatest workload burden, while 83 per cent had adopted ambient tools before their organisations introduced formal policies. Those figures are drawn from the company’s own survey, rather than an independent assessment. If the second finding reflects the wider profession, it suggests that clinical enthusiasm is running ahead of institutional governance. That should concern NHS leaders as much as it interests technology suppliers.

There is another question that tends to disappear behind claims about administrative efficiency: what happens to the time saved? Giving a doctor fewer minutes at a keyboard does not automatically create additional clinical capacity. Someone must decide how that time will be used. In Seoul, the pressure of exceptionally high patient volumes makes it easy to imagine hospitals converting saved minutes into additional appointments. In England, the decision is more politically charged. Trusts facing financial deficits and lengthy waiting lists will naturally see an opportunity to increase throughput. Clinicians already working under considerable pressure may regard the same time as an opportunity to complete their work without routinely extending their working day. Both expectations are understandable, but they represent different definitions of success. Without an agreed purpose, NHS organisations will struggle to establish whether the technology has delivered what they paid for.

For NHS leaders, the immediate responsibility is to demand evidence that goes beyond demonstrations of impressive transcription. Suppliers should be required to show how accurately their systems capture clinical information, how much time doctors spend reviewing and correcting generated notes, and how they perform in comparable healthcare settings. Contracts must specify where audio recordings and transcripts are stored, who can access them and how the software interacts with existing records. The regulator’s July clarification also matters because it distinguishes functions that fall within medical device legislation from those that do not. Procurement agreements should reflect the regulatory position of each feature rather than treating every ambient AI product as interchangeable.

Policymakers have an opportunity to address the wider problem. NHS England’s supplier registry could evolve beyond a directory of self-certified companies into a clearer set of national expectations covering clinical note structures and electronic record integration. Such guidance would give individual trusts a stronger starting position when negotiating with suppliers. It would also make interoperability a condition of wider adoption rather than an issue left to resolve after contracts have been signed. For companies operating in Britain, including Heidi through its London base, participation in this emerging standard-setting process could become an important determinant of their place in the NHS market.

If Samsung Medical Center’s pilot disappoints, the memorandum may quickly disappear from view. The strategy behind it, however, deserves attention. Agree the standard with a prominent first customer, establish a working reference model, then take that arrangement to the wider market. England is following a different sequence. Adoption is gathering momentum while the standards governing the clinical note are still being shaped by local contracts and individual supplier relationships. There is time to change direction, but the window will narrow as more trusts commit to particular systems. Once those arrangements become embedded, the clinical note used across much of the NHS may reflect the commercial preferences of the supplier that secured the earliest and largest foothold, rather than a standard the health service deliberately chose.