From Sustainability Assessment to Decision Records: The SEEDS Framework for Sustainable Requirements Engineering in Digital Health Software

In the modern world, software is no longer just a tool; it is the invisible infrastructure of our daily lives, from the apps that manage our bank accounts to the systems that coordinate emergency care. When we build these digital systems, we often ask if they work well, if they are secure, and if they are easy to use. However, a growing realization among engineers is that these questions are not enough. A truly successful system must also be sustainable. In this context, sustainability does not simply mean saving electricity or reducing carbon footprints, though those are parts of it. It is a much broader concept that asks whether a piece of software can last for years without breaking, whether it treats all users fairly, whether it fits within a hospital's budget, and whether it supports the mental well-being of the people who use it every day. The challenge has always been how to turn these big, abstract ideas into concrete instructions for the people who write the code. Without a clear path from a vague goal like "be more sustainable" to a specific rule for a computer program, these goals often remain just wishes, never actually built into the final product.

A team of researchers from the Federal University of Health Sciences in Porto Alegre, Brazil, has developed a new way to bridge this gap, specifically for the complex world of digital health. They created a framework called SEEDS, which stands for a structured approach to making software that is sustainable across five key areas: how it affects the individual user, its impact on society, its economic viability, its technical strength, and its environmental footprint. The researchers recognized that a patient, a doctor, and a software developer all see the same piece of software in completely different ways. A patient might worry about whether an app is easy to read, while a developer worries about whether the system can handle millions of users without crashing. The old ways of checking for sustainability often treated everyone the same, asking patients to judge technical details they could not see, or asking developers to guess how a patient feels. This new framework fixes that by insisting that every concern be viewed through the specific eyes of the person raising it.

The heart of this new method is a document called a Sustainability Decision Record. Think of this as a detailed logbook that connects a problem to a solution. When a stakeholder, such as a nurse or a hospital manager, identifies a sustainability issue, the framework guides them to write it down in a specific format. This format forces them to define exactly who is seeing the problem, what part of the five sustainability dimensions it touches, and what specific rule the software must follow to fix it. For example, if a patient says a health portal is too confusing, the record does not just say "make it better." It translates that into a specific requirement, such as "the system must use simple language for all medical terms," and then lists exactly how to prove it works, like running a test with real users. Crucially, the record also forces the team to write down any trade-offs. If making the system simpler for patients makes it harder to keep secure, that tension is written down explicitly, ensuring that no one forgets the cost of their decisions later on.

The researchers tested this approach in three very different situations to see if it worked in the real world. First, they looked at digital health systems that were already being used, asking how the framework could help improve them. They found that by looking through the lens of specific roles, they could spot hidden problems, such as a system that was technically sound but confusing for elderly patients, and turn those observations into clear instructions for updates. Second, they applied it to the process of deciding whether to adopt a new digital health solution for a public health system. Here, the framework helped managers ask the right questions before signing a contract, such as demanding proof that a new tool could share data with existing systems, preventing the future problem of being locked into a single, expensive vendor. Third, they used it when a hospital was asking for a brand new system to be built. By introducing sustainability rules at the very beginning, they ensured that issues like data privacy and long-term maintenance costs were considered before a single line of code was written, rather than trying to fix them after the fact.

To address the fact that writing these detailed records can be time-consuming, the team also explored using artificial intelligence to help. They built a system where a large language model could read the initial concerns and draft the first version of the decision record. However, they were very careful to design this so the computer acted only as a helper, not a decision-maker. The artificial intelligence could suggest what the rules should be and what evidence was needed, but a human expert had to review every single suggestion, correct any mistakes, and make the final call. This approach showed that while computers could speed up the paperwork, the human judgment of doctors, nurses, and managers remained essential to ensure the decisions were safe and fair.

The study suggests that this method successfully turns the vague idea of sustainability into something engineers can actually build and test. By making the connection between a person's worry and a specific software rule, the framework ensures that sustainability is not just a buzzword but a traceable part of the software's life. The researchers found that when teams used this method, they were better able to see the conflicts between different goals, such as the tension between making a system fast and making it secure, and they could document their choices clearly. This means that years later, if the software needs to be changed, the next team will know exactly why the original decisions were made. While the researchers note that more testing is needed to see if this works in other countries and for other types of software, their work offers a promising path forward. It shows that for digital health to truly serve people for the long term, the people who build it must have a clear, shared language for talking about what matters most.