Design in the data-driven future of healthcare
Last update: April 2024
How thoughtful design can help make good on the promise of data in healthcare.
For the past year, COVID-19 dominated the headlines. Together with all the news articles covering the pandemic came a broad range of data visualizations, graphs, and charts that aimed to make sense of it all. This coverage highlighted the importance of good Data Design in explaining healthcare topics, which by their nature are complex. Not only with COVID-19, but with healthcare in general, which is becoming more and more data-driven.
This article explores some of the opportunities and challenges of Data Design for healthcare applications, as well as some practical considerations. In future posts, we aim to explore more specific applications and examples from our projects.
Healthcare and data
Healthcare, data, and visualization, have been closely connected throughout history. For example, the novel visualizations that Florence Nightingale created to illustrate seasonal sources of patient mortality during the Crimean War in the 19th century. Or how John Snow created dot maps, to investigate a London Cholera outbreak around the same time. In both cases, Data Design helped save lives.
As the role of data in healthcare evolves, so does the amount of data that professionals need to view and consider when making decisions. Furthermore, the audience is growing beyond healthcare professionals. This makes it all the more important that tools and applications are designed well. Ultimately, the goal is not more data, but better insights and better health outcomes.
Opportunities for Data Design in healthcare
Within the broad domain of healthcare, we see a number of areas of opportunity where thoughtful Data Design can add significant value.
1. Public health
The previously mentioned COVID-19 dashboards are a great example of how making health data public can benefit all. Similar open data dashboards have been around for some time, created by cities, governments, or public organizations. Open access to data allows researchers, organizations, and the public to gather insights and set strategies to improve public health.
To benefit those without data analysis experience, this means that there must be easy-to-use tools available to explore the data. Often this is not the case. Either because it is overlooked or the budget is not made available. As a result, the potential of the data remains hidden behind complex language and obscure interfaces. And therefore limiting the use to only a small set of professionals. A better design would ensure a broader reach, increased ROI, and more support for these tools.
Examples
- Our World in Data — Open data on global topics, including health.
- NY Times COVID-19 Data — Example of well designed visualizations tracking the pandemic.
- The Official Corona Dashboard of the Netherlands by CLEVER°FRANKE — The central hub of information and up-to-date data from the Dutch government during the COVID-19 pandemic in the Netherlands.
- City Health Dashboard — Explore and compare public health data for more than 500 cities across the US.
2. Machine Learning and Precision Medicine
Combining large data sets with Machine Learning and Artificial Intelligence brings new possibilities of better identification of health conditions, such as cancer. With Precision Medicine, a patient is matched with the optimal treatment based on their unique genetic and physiological profile. An ever-growing data set of patients who were treated for the same condition forms the basis for these. While receiving the recommended treatment, every patient is, in turn, contributing data back. Each patient helps refine treatments for others.
Step by step these applications are inching closer to the accuracy of human experts, and in some instances already surpassing it. The design of these tools brings the crucial challenge of having professionals understand and trust the conclusions by these systems. Results must be presented in a way that helps specialists navigate large amounts of data, but also help patients understand relevant results in a simple way.
Examples
- Tempus ONE— Voice-enabled device provides physicians real-time access to Tempus reports.
- Parkview Medical Patient Satisfaction Platform by CLEVER°FRANKE— Platform to track and act on patient needs and satisfaction throughout their medical journeys.
- Evidation Health — By applying behavior data outside clinical settings Evidation creates new research opportunities and insights.
- SkinVision — Detect and track potential skincancer spots with a phone app.
3. Connected Personal Health
Outside the clinical setting, personal health is becoming more data-driven too. Where the Quantified Self was once a fad, it is now mainstream with activity trackers, smartwatches, insulin sensors, and such, continuously gathering data through an easy-to-use form factor. As sensors become more capable, new opportunities come with. From basic heart rate, we’ve moved to EKG analytics that can detect heart conditions. And continuous blood pressure and blood levels are on the horizon.
These applications create a continuous feedback loop that helps the users (and healthcare professionals) optimize their health. Good Data Design is needed to help interpret the data and place it in perspective for the user. Keeping in mind that the user is likely not a medical expert.
While these devices point to a future with more tailored data-driven personal health guidance, there are still important limitations. For example Apple Watch only can detect atrial fibrillation. CNET provides perspective on its usefulness.
Examples
- Apple Watch Heart Monitoring — How Apple Watch helped detect a potentially fatal heart condition.
- Dexcom G6 — Continuous glucose monitoring helps patients better manage their diabetes, with improved health outcomes.
- Patient Support App by CLEVER°FRANKE— With an approachable and friendly design, this app uses feedback data to provide personalized exercises to help patients cope with cancer-related fatigue.
- Pictal Health — Helps patients document their condition and data visually to better communicate with healthcare professionals.
4. Electronic Health Records
More and more the medical history of a patient comes together in an Electronic Health Record (EHR). This includes current and previous treatments, laboratory tests and more, from a single or multiple healthcare providers. Ideally, this improves the accuracy and efficiency of care, as well as reduces the risk of medical errors. For example by checking prescriptions for interactions with other medication a patient is taking.
However, the diversity of information stored in the records and the flexibility needed to manage complex health conditions brings complexity. Many systems accessing these records have elaborate interfaces that are difficult to navigate. The result is that they add to the workload of already time-pressed healthcare workers. A 2017 study concluded that on average a US doctor spends almost 6 hours a day using an EHR system. So any inefficiency, frustration or inaccuracy adds up significantly. A human-centered Data Design approach can help in creating systems with more clarity, and that provide the right information at the right time.
Death by a Thousand Clicks — Fortune Magazine together with Kaiser Health News did a joint investigation into the history and issues with Electronic Health Records.
Examples
- HMF OncoAct by CLEVER°FRANKE — By redesigning their patient report, we made treatment details more accessible to physicians and patients.
- Inspired EHRs: Designing for Clinicians — Open source book on designing usable Electronic Health Records with example designs and clinical scenario’s for reference.
Challenges
So, what makes designing with data particularly challenging and important in health applications?
To start, our bodies are complex systems, and our health is not defined by a single clear metric. Rather, it concerns a wide range of data points. Often over a longer period.
The data itself is fuzzy. Some measurements are difficult to take accurately. Each individual is different and responds uniquely to medical treatment. What is a good value for one, might be poor for another. Values also can vary greatly with the time of day and situation. Together this means that data must be presented in perspective, with flexibility and the right reference points.
A wide range of people is involved. Not only clinicians and patients, but also nurses, specialists, family and significant others. Each has different information needs, levels of knowledge and emotions. For medical professionals, there is the added factor of time pressure.
In the end, we must get this right. Healthcare decisions have a huge impact on a life, now or later. It does not get more personal than this.
Designing with healthcare data
Success in design starts with understanding the problem in detail. In particular with healthcare applications with all their complexity. Based on our experience with Data Design in healthcare, these are some considerations to keep in mind.
1. Start with the user in mind
First of all, it all starts with the user(s) and how your product fits in their life. In particular with healthcare, it involves a diverse range. From patients and clinicians, to caretakers, specialists, nurses, insurance specialists, the general public, researchers, etc. Each has a different level of expertise, information need, and different emotional connection with the data and any conclusion. To surface the first inventory of this in our design process, we use methods like Moments that Matter.
2. Co-design with the specialists
To build on the previous point, make specialists an integral part of your design process. They have a deeper understanding of all the nuances of the topic and data than a designer can ever hope to grasp in a design project. And they probably already have smart ideas about how to tackle certain (design) challenges. So leverage that by bringing them in, sketching together with them and, validating your designs. Together you can come to better solutions faster.
3. Consider the data
Then consider the data and its character. This defines how it is best presented. For example how quickly values are expected to change. Or what fluctuations are to be expected around a trend. A blood pressure measurement can vary greatly depending on the time of day, setting, or previous activity, but long-term trends are what often matters most. Visualizations should provide the right perspective.
4. Don’t alarm
Similarly, pay attention to finding the right level of urgency in presenting data and results. For patients, who have no medical degree and who view everything from their personal perspective, an adverse trend, message or data point can be a source for great concern. In particular this is the case when presenting risk levels or comparing results to the general population.
5. Make it actionable
To the extent it is possible, aim to make the data actionable. Particularly for non-experts, it is important to know what to do, or not to do, based on the insights from the data. Provide context to help the user understand the values and, over time, gain an understanding of how their behavior might affect them. This also helps build trust in the data and following actions. Keep in mind to what extent providing advice is (legally) appropriate. Even if advice within the application is not an option, having a line of communication to a healthcare profession, who can advise, can be of great help.
6. Keep it simple
Finally, the greatest challenge with the previous considerations in mind; keep it simple. The goal is to add insight and not just add data. Keep the focus on the core goals and values. Eliminate as much as you can, and give everything else the right place and priority.
In closing
We’re at the threshold of a data-driven healthcare revolution. Compared to the time of Florence Nightingale and John Snow, data has become richer, real-time and more insightful. And we’re just scratching the surface. As we shift to personalized healthcare, data will be viewed and explored by all of us. To navigate the complexity and, make good on the promise, we’ll need good design.
In future articles, we’ll explore specific healthcare topics, and what design methods you might use.