A glossary: To be able to use data in a smart and efficient way, it is not always possible to keep doing things the way they have always been done. For example, computers need help to understand us, and that is why a ‘translation’ may be helpful. Within Swelife’s Sweper project, a group of urologists have worked specifically on this: on ‘translating’ technical terms relating to urinary bladder cancer into an international code system. This ensures that systematised information can benefit healthcare.
“We have produced a selection from an international code system, Snomed CT”, says Anna Rossander. “The terms are now in Swedish and matched to a national Swedish register.”
Anna Rossander is a specialist in urology and doctoral student in informatics at Applied IT, University of Gothenburg, and has worked on the project together with Truls Gårdmark, a urologist and senior lecturer at Karolinska Institutet. Ulf Lönquist – INCA IT coordinator at the Regional Cancer Centre – was the project manager.
The register used was the Swedish National Register of Urinary Bladder Cancer. The project produced a machine-readable code system of relevant urinary bladder cancer terms, like a glossary. Now, the glossary makes it easier to introduce the code system Snomed CT in other systems, which facilitates information sharing, so-called single entry (instead of duplicate entries).
If, for example, the medical records system uses different terms for different things, the computer cannot easily ‘understand’. It is important to create non-ambiguity. Since the information is then possible for a computer to process, there is an increased opportunity to produce and implement tools such as automated decision support, machine learning, more advanced data processing and artificial intelligence.
Automating information
“Since 1997, we have tried to understand how urinary bladder cancer works for the patients”, says Truls Gårdmark. “It is important that the variables and information are correct in the medical records system. When we have coded it, the information can be sent on automatically instead of manually as it is today.”
“Single entry is an advantage in structured data”, says Anna Rossander. “Then there are other advantages: structured data is easier to analyse, link to decision support and conduct research on. Snomed CT also has a powerful built-in logic that makes it possible to request and see connections in the information in a different way to conventional code systems.”
“This makes you better prepared when you are getting started with artificial intelligence and machine learning”, says Ulf Lönquist.
Carried out by professionals
One of the great things about the project is that it is carried out by professionals. Both Anna Rossander and Truls Gårdmark work as urologists and have good knowledge of how work with urinary bladder cancer takes place in practice.
“We know the domain and have based our work on what is relevant to healthcare”, says Anna Rossander. “This is the information that the profession thinks is most important for these specific patients. We have used medical language that is possible to share internationally.”
“If it is possible to do this automatically, it opens the way for seeing the processes, and enables an overview”, says Truls. “The work not only provides an opportunity in the medical record keeping to collect the information in a structured way and be able to use it more efficiently – it also provides new insight into how the register itself is constructed. We see inconsistencies and other phenomena that can occur in the creation and maintenance. The register could be structured in a smarter way! Since I am the keeper of the register, I can address this and correct it directly together with the support team, which is really efficient.”
Method book
In addition to producing the glossary, the project also includes a method book for the project delivery. The group has documented the process of producing the glossary in order to develop the method for mapping in other areas. This took place in collaboration with the National Board of Health and Welfare. The method book will shortly be available for download from the authority’s website, socialstyrelsen.se. It is designed so that others are able to use it, assuming that they have a certain level of understanding of how Snomed CT works.
“It is best practice”, says Ulf Lönquist. “This is quite difficult, so one shouldn’t trivialise it. It takes time and is relatively complex. The underlying structures are complex polyhierarchies.”
“The key was that we decided that this was a process”, says Truls Gårdmark. “There is a generic model for how to design a process, with project managers, reporting etc. We have followed that model and added to it as needed.”
Possible to reuse
“Reusability may be a buzzword, but here it is truly reusable in many different ways”, says Ulf Lönquist. “If you are working internationally this is very well established, and you obtain the level of detail that you require.”
“There are certain domains of these registrations that others can benefit from”, says Truls Gårdmark. “For example, we have met with oncologists, and they have regimen libraries in which the treatment of different forms of cancer are to be found; some are used in urinary bladder cancer. This makes it smarter to use the same registration method that we use when we register urinary bladder cancer, something no one had thought of previously. If everyone is able to use the method, the benefits increase. In creating the glossary, we visualise problem areas that are yet to be seen. It has initiated improvement work. So the project has a reinforcement effect.”
“I work with the patient overviews [for cancer care, which is also a Sweper project], and these are connected”, says Anna Rossander. “A lot of what we have done in urinary bladder cancer can be used in the patient overviews. The National Board of Health and Welfare has been issued a new public service agreement including coordination responsibility for cancer. Therefore, they are also keeping their eyes on the ball, and we hope to be able to reuse both our methodology and the glossary. We cannot predict where the ball is then going to roll, but at least it will begin to roll.”
“You have to start somewhere”, says Ulf Lönquist.