IMEDALytics: Healthcare Machine Learning – Part 2

Annika Kaltenhauser Senior UX Designer • Senior UX Designer

Dr. Verena Rheinstädter Senior UX Researcher • Solution Expert User Research & Qualitative Methods

22/02/2021 • 7 minutes reading time

How can designers help improve the care of patients in intensive care units? This is exactly the question we are addressing in the IMEDALytics research project.

In this article, we cover the significance of three context analyses that we carried out as part of this project — as well as the valuable knowledge we gained. Find out all the details surrounding problem assessment and project objectives in the first part of the article.

Understanding the system’s use context

So that we can guarantee a good user experience, we have to develop a comprehensive understanding of the intended user and the future system’s technical possibilities.

Contextual inquiries

After a scoping workshop with all project partners, we paid three visits to intensive care units at the University Hospital of Aachen and Dortmund Hospital. We were able to observe the everyday work across various intensive treatment units as well as accessing the medical expertise of the staff there — also our future users.

Contextual inquiries combine observations and interviews and are an essential part of this project for us as UX Designers. They offer us the unique opportunity to identify user contexts, users and important factors impacting healthcare professionals’ decisions and actions in their stressful working environments.

As well as this insight into real everyday work, the visits let us identify how we can integrate IMEDALytics into medical and nursing staff’s everyday life at the hospital so that the system actually provides support.

Learnings

1 Handovers

It was enlightening to observe that there are informal and formal information handovers as well as synchronized and asynchronized collaborations between members of staff. Shift handovers take place between

  • doctors and other doctors

  • nurses and doctors and

  • nurses and other nurses

Additional consultations are spontaneous — particularly when a second opinion is required. During examinations, yet more information is exchanged and decisions made as a joint effort.

2 Role sharing

While doctors see their patients at relatively long intervals and decide the appropriate therapy, nursing staff see their patients more often so are responsible for monitoring their status as well as achieving targets. Their observations on patient condition and the success of assigned measures form part of the doctors’ decision-making processes.

3 Documentation: digital and analogue

During our visits to the two hospitals, we found out that the units we visited differed drastically in how information is documented using digital systems. Doctor’s orders, prescribed medications, treatments and other care instructions were simultaneously logged by doctors and nurses. In one clinic, this documentation was digital, while the other used standardized sheets of paper — this makes a major difference in the context of our project!

4 Workload and stress

Staff workloads deviate between weekdays as well as during the course of one day, depending on how many patients are in the unit and how critical their conditions are. Then there are factors such as a constantly high level of noise from medical devices. Many generate acoustic feedback even during less critical events. In addition to this stress, staff had to make repeated, sometimes lengthy trips between individual rooms and were rarely able to take breaks.

User experience map of two nursing staff
User Experience Map

Findings are established in personas

In order to condense the knowledge we gained as well as develop and contextualize important user characteristics and requirements, we created multiple personas. Personas illustrate hypothetical people to represent typical users. The generated personas are involved time and again throughout the course of the project. This lets us ensure that prospective users actually remain our focus. What’s more, personas facilitate targeted communication between all those involved in the project.

In some projects, creating just one persona is sufficient. Within the context of IMEDALytics, we differentiated between three groups that we categorized based on their different needs regarding the system:

  • Regular unit doctors

  • Nursing staff

  • Researching doctors

With the help of our personas, we are able to highlight how their needs differ.

Persona: Hanna Tolle, assistant doctor
Persona: Hanna Tolle, assistant doctor
Persona: Peter Busch, nursing specialist
Persona: Peter Busch, nursing specialist

Impacts on our design

One desire very clearly crystalized during our contextual inquiries: Nurses and doctors want a decision support system (DSS) to support indication, risk assessment, choice of the right therapy and monitoring as well as therapy management. Furthermore, doctors in particular stressed how very important it is to make a system’s decisions transparent and easy to follow.

“That would be a dream come true” (unit doctor).

The various sub-goals for future users resulting from the treatment phase and role (decision-making, implementation, review) form the framework for our DSS design.

Other than this, we came across contrasts that impact our design mission:

  1. A future system should combine the intelligence of machines and humans to guarantee the best possible care. An interactive system must also bring together the knowledge and skills of various people in a way that makes sense.

  2. Requirements and targets are greatly defined by the current context; while it is sometimes important to display decisions transparently and in a way that is easy to follow, other situations may make it important to portray the decision as efficiently as possible to avoid stress-related errors.

To sum it up

For meaningful, efficient UX design, it is essential to speak with future users and observe them in their everyday work environments. They will come up against real challenges and stressors that impact the users significantly.

Only when a design approach takes this into account — and provides a solution — can it be perceived as helpful. Especially when dealing with an extremely complex system such as IMEDALytics, presentation must comply with user needs for transparency and ease of understanding.

Logo of the Bundesministerium für Bildung und Forschung

Addition: The project on which this report is based was funded by the Federal Ministry of Education and Research under the reference number 13GW0280 B. The project is funded by the Federal Ministry of Education and Research. Responsibility for the content of this publication lies with the authors.