“This is what it really comes down to: how do we reduce the administrative burden and create greater peace of mind?”
— Femke van Wetten, Healthcare Applications Manager at GGZ Oost Brabant

Background

GGZ Oost Brabant has around 2,000 staff and volunteers working across several locations in the eastern part of North Brabant in the Netherlands. They provide specialist mental healthcare to clients of all ages, from children and young people to adults and older people, with complex mental health difficulties such as anxiety and mood disorders, trauma and personality-related difficulties.

Being an attractive employer matters to GGZ Oost Brabant. The organisation wants to be a place where people enjoy their work. In practical terms, this means ending the working day with peace of mind and a sense of having completed what needed to be done.

The challenge

Within the organisation, healthcare professionals, from nurses and support workers to psychologists, healthcare psychologists and clinical psychologists, spend too much time on indirect tasks rather than direct client care. According to Femke, this affects the very core of how healthcare professionals experience their work.

“Our healthcare professionals are here first and foremost to treat and support our clients. That is why we need to take a critical look at our administrative processes. By using AI as a supporting tool, we significantly reduce the administrative burden and give our healthcare professionals more time for our clients and their care needs,” Femke explains.

Time-consuming tasks affect not only the time available for clients, but also job satisfaction and workload, and can contribute to sickness absence and staff turnover. Replacing staff is particularly difficult in the region served by GGZ Oost Brabant: according to Femke, the available pool of people in its catchment area is roughly one-tenth the size of that in a comparable area in the Randstad, the densely populated western part of the Netherlands. This is not a problem that can simply be solved by stepping up recruitment.

The more colleagues Femke speaks to, the clearer it becomes what these indirect tasks involve:

“The more colleagues I speak to, the more I learn that a large part of it involves drafting reports and letters. They do not yet realise everything that AI will make possible. Ultimately, we want Delphyr to support us throughout the entire treatment process,” Femke explains.

Another issue is that qualitative and quantitative information in the record is currently not captured in a sufficiently structured way. Healthcare professionals mainly make decisions based on their knowledge and experience, even though structured data can be highly valuable when comparing a case with others. According to Femke, Delphyr could eventually contribute here too by supporting decision-making.

Femke stresses that using Delphyr to support treatment reviews with structured data and insights from previous or other cases is still a future ambition. It is, however, a clear strategic goal for the longer term.

Guidelines are another consideration: healthcare professionals currently have to visit separate websites to access the GGZ Standaarden, the Dutch mental healthcare standards, national guidelines or the Farmacotherapeutisch Kompas, the Dutch reference for pharmacotherapy. This takes time and interrupts their workflow. With Delphyr, these guidelines will become directly accessible from the record itself.

The solution

Delphyr supports healthcare professionals at GGZ Oost Brabant in three areas: finding and summarising information more quickly, accessing guidelines directly and automating time-consuming workflows.

Search & Summarize

Healthcare professionals currently search records manually and compile summaries by combining passages of text. Femke expects Search & Summarize to have the greatest impact on their day-to-day work, particularly when combined with the speech-to-text solution the organisation already uses.

Guidelines

Delphyr is connected to the national guidance that GGZ Oost Brabant trusts: the GGZ Standaarden, national guidelines and the Farmacotherapeutisch Kompas. Femke demonstrated this live at the NVvP conference of the Dutch Psychiatric Association. In her view, it already worked well: combining guidelines with EHR context produced a specific, evidence-based answer to a prescribing question, including points to consider based on protocols and the client’s medical history.

An important additional benefit is that Delphyr only queries trusted sources connected in advance, keeping the risk of incorrect or fabricated information lower than with a general-purpose AI model.

Automatically generating a first draft

Alongside searching, summarising and consulting guidelines, Delphyr will also help healthcare professionals draft recurring documents. Instead of writing a report or letter from start to finish themselves, Delphyr will automatically generate a first draft, which the treating clinician reviews and refines as needed.

Three workflows are first on the roadmap: preparing for consultations with the clinician responsible for coordinating treatment, preparing for multidisciplinary team meetings and drafting discharge letters. These are not yet standalone features, but part of a broader ambition to use Delphyr throughout the treatment process, from intake to discharge, in line with GGZ Oost Brabant’s vision of care.

The organisation is also working with Delphyr to improve data quality. By writing words or sections of summaries directly to the appropriate Dutch healthcare information building blocks, known as ZIBs, data can be structured and exchanged more easily between healthcare organisations. Femke sees this as another step that still needs to be developed, but one with clear and considerable potential.

The partnership so far and expectations

As GGZ Oost Brabant is still in the process of rolling out Delphyr, this section focuses on the journey so far and expectations for the coming months, rather than results already achieved.

Femke is clear about the partnership with Delphyr so far: “Two thumbs up.” What has stood out to her is that the Delphyr team has proactively addressed privacy and compliance from the outset, including the DPIA (Data Protection Impact Assessment) and a risk assessment covering the availability, integrity and confidentiality of data, known in Dutch as a BIV risk assessment.

In terms of expected impact, Femke believes that making it easy to create summaries with Search & Summarize, combined with speech-to-text, will have the greatest immediate impact on healthcare professionals. She expects the other steps primarily to improve the quality of care, with their effects on healthcare professionals’ daily experience becoming apparent more gradually:

“That will happen step by step, because it also requires a change in mindset: healthcare professionals need to learn to trust the advice they receive, while remembering that decisions remain their own responsibility.”
— Femke van Wetten, Healthcare Applications Manager

Femke has clear advice for colleagues at other mental health organisations who are unsure about starting a similar project: “Just do it. Do not be afraid. Start with the simple things. Search & Summarize is not necessarily very high-risk, because it does something you have been asking for all along: produce a summary. It is mainly about having the courage to take that step.”

What’s next

The partnership will enter the testing phase over the coming months, before the rollout to healthcare professionals begins. During implementation, GGZ Oost Brabant will train staff team by team, with team leaders scheduling training within existing meeting slots. The goal is for everyone to have completed training before the end of the year.

This approach is based directly on a lesson from an earlier, similar implementation: training teams together works well. Teams often include colleagues with varying levels of digital proficiency, and training them together enables them to continue turning to one another for support.

GGZ Oost Brabant will assess success primarily in qualitative terms: a reduction in sickness absence and the findings of a staff evaluation centred on a simple question: “Do I have peace of mind?” Quantitative results are also expected over time, but Femke believes these should emerge from care delivery itself and do not need to be imposed from the top down.