4#9 - Marte Kjelvik & Jørgen Brenne - Healthcare Data Management: Towards Standardization and Integration (Nor)

That's the Metadama, a healing podcast on data management in Noden. Hi and welcome. I'm Winsrit and thanks for joining us on a new episode of the Nodens Podcast, where we show you the data management in Noden at Løv. We show you the future of competence in the field of artificial intelligence. Therefore, we invite you to join me as a Nodens expert in data management and information management at Løv. Welcome back to Metadama. As always, I'm super happy to have a new episode of the Nodens Podcast. Today, we're going to talk about health data. My name is Marte Kjellvik and Jörgen Brenner, both of whom work at FHI, the People's Health Institute. I heard about the presentation both of you had in the end of April, the 24th, in Stockholm at the Data Innovation Summit. It's a topic that I think is very exciting, not only for health data, but also for many others. And that is how you integrate, how you work with both of these fine, titular presentations in revolutionizing health data integration. So, very exciting to have you with me today. Welcome. Thank you very much. Integration and interoperability are perhaps some of the areas that are continuously honest during development, and it happens a lot. Not least, since Gen I revolutionized health data, but integration techniques have changed so much and are actually continuous. At the same time, it is now the most fundamental that we have in the way that we organize around data in organizations. And the balance between the introduction of new technologies, developing on one side, on the other, building up a foundation to stand on, is not always as easy. And then there are many suppliers on the market, and that might be the market for data, where there are mostly suppliers, instead, which also have influence on which direction the market is developing. How do you manage to keep an eye on this, and what is actually needed? And that is perhaps some of the topics we are going to talk about today. But before that, Marte and Jörgen, you can introduce yourselves. We will start with Marte. First of all, thank you very much for the invitation. I am very happy to have been part of this podcast and learned a lot from what others have also been doing. My name is Marte Kjeldrik. I am a project manager at FHI in Avdeling, or Klinger for Pistra. We are working on a big modernization of the registry. We have done a lot of work on the development of new areas, so that is what the project manager is doing and what we are helping with. My name is Jörgen. I am an engineer as a technical director in the project. I have been working with data for 18 years. The main thing is that I am very interested in data security, so that I use data as a source of information, and I am focused on that. Very exciting. What is it that you do when you do not work with data security? Perhaps you, Marte, are responsible for that. I can also say that it is a demanding job that we are working with. It has been a project leader. It is both very difficult and the management of communication at different levels. It is a lot that happens, and it is hectic. For my part, I am happy to be able to find something else when I am not at work. It is training and dancing samba at Africa. It is something that I am very happy with. Do you want to talk a little bit about where you started to get into the field of interest and where data comes from? My part has started as an analyst at NPR and KPR, a Norwegian patient register. I came up as a patient and user register in 2007. I have worked with data security for many years as an analyst. I have also been asked in 2019 to be productive for data security, and then came into a world that was very interesting to use methodology and work height to standardize and also saw that the importance of data quality, that you standardize and use methodology and saw that challenges were actually in the forefront of data security. I can ask you, with your experience and background, do you think that standardization within the data is at the same level as in other areas, or do you think that we hang a little back when it is standardization? There is no big challenge for health data. It is also a well-known challenge that many jobs and many actors have to share. For example, when you compare data on the opposite side of the field that you are going to do, there are many different registers that you are going to compare with other types of registers. If you have not grouped them in the same code, for example, then it is hopefully the same style of health data. You have a long way to go, and that is something you work with on the opposite side of more units and actors. I have previously said several times that a desire with this podcast to try to balance, that regardless of how much experience you have in the field of knowledge, you should be able to take part in what is out of our list on this episode. That is why I am very happy to have both Marte here and Jörgen. We can have both the big frames in place, but also dive a little into the technical aspects. I think that you have already started in the direction of the big frames. Marte, maybe you can explain a little more about the people's institute, what are the challenges you see as an intro to their situation. First of all, since January 1, 2024, we have been a health register part of the People's Health Institute. Previously, we were in the Health Directorate, and we have been through a task from the Health Administration Department, to collect the register in the People's Health Institute, which is now about ten registers. And it is an important way to start with the standardization and the summary, and ultimately the win, not only for the register, but also for the service out there, which may be driven by some double registration. There are different demands from different actors who are very involved in the organization, but also working on a strategy in the People's Health Institute, on how to collect the register in the future, and what is the goal setting for the future. So it is a challenge that we... There are many registrations and differences in the level at which a degree is modernized, and some have a questionnaire, while we have automated it in a different way. But we also need to modernize ourselves, because we need to have more data, and we need to have more information. We did not see this through the pandemic, that there was fresh data, to follow how many people were involved with Covid, and we need good management data to be able to plan the services in the future. I think at the starting point, we had a previous episode with the Ministry of Finance, where we talked a little about the registration of women. It is a huge advantage to have good registration data, but also demanding, on the one hand, to collect data, but on the other hand, to keep the information safe in a good way. What do you see the biggest challenges with the registers you are in, and what do you expect? The biggest challenge, the most important thing to take into account, is the standard at the national level. The first thing to understand is, for example, what the health service is, and what services you have in the health service. How do you define that? It is a matter of understanding how you can take data from which services. Also, you have code works that are standardized, which are implemented in the different electronic patient tunnels, such as the hospital, the community and so on. The service has a system, and there are many systems today, that need standardization, but it is again standardizing requirements from the Ministry of Finance and Health. Also, it is to be close to the health service and the good dialogue in the health service, and what is actually the purpose of the data, close to the work process in the health service. The national part is to be able to get good national numbers and collect data across the unit, across the area, and follow the patient through life and through various services. It is very important to get good data on the site. Now you are lucky to have Jagen with you as a technical project leader. But what do you think of the over-earning levels, to open up the goals you have set, what skills and competencies you need? No, here we build our competencies parallel while we go, and what we see is that the organization needs to build competencies on data governance, data management, so we do that in parallel. We have been doing that for half a year, for a year and a half back. So we took quite a broad approach in the organization, that 5 out of 40 got basic course in dimension modeling, for example with Margueross from the USA, and then we are able to get to the field language, to understand each other across the area, between both technical and technical, as we can do in the organization. So it is important to have a field understanding around what this is from something, and it is very important for the competence that everyone gets to learn data management and data management. What I find exciting is that there is a quite extensive focus on data modeling in their work. And it is, it is difficult, it is difficult to believe that you have that focus. It is also because in the last 10 years, from my perspective, I have built in data modeling in the applications, that it is very likely to be a door-based business, and then you miss a little the summarized focus on what it actually means in the framework and in the context of my company and my organization. So you have tried to identify these problems, perhaps in a strategic way, and an actual architecture perspective. So maybe we will win a little strategic, and talk about how you have thought strategic on organizing data, on the other hand, at least standardizing. Maybe you can tell Marta a little about how a strategy can be developed. Yes, it just looks like that. We are working on a national strategy, which is important first of all, and that we have an actual strategy, and that is to improve the management. It is very important. It comes from the management, that we are going in this direction, and that we do not have any direction to go. And then we have established a work-based process, how to work with new data sets. And then we have established that we are going to start with the user history first. And then we will prioritize the user history. It is not an advantageous user history, on a technical level. But for example, under the pandemic, the weight of young children is increased. So a user history on a data set that we are going into in the North-East region of the school district, is separate children with children. And what is it that we need from the understanding, and data to be able to respond to user history. So let's start with user history. Then we can move on to conceptual models. In this context, we are talking about a model for the understanding of the management. What are the concepts that define the field of management and the management of the area. And then the health services. As we usually say, it is 80% of the youth. It is there, in the work, to describe the understanding well, and the relationship between them. Before we have logical models and data contracts, with the transport from the EPJ, also from the system out there and into us. And the whole thing is that the course is modeled in the register, with the methodology we have chosen there. And also that the course data is used. Now I use Power BI. The course structure is that one can use user history. So in the work process, we have defined the strategy for how we are going to work on new changes in the data set. I think this is very exciting when you talk about user history for conceptual models, and connect this very closely to the understanding of management. You might say something about how this looks like, in terms of technology. How does it work? The conceptual models are very focused on defining the subject of management and the relationship between them to identify that it is not a technical model. And these models are very important to be able to discuss with all different actors. We have legal requirements. We have to talk about legalists. We are talking with professionals who do not have legal skills. And also the conceptual models can be used to develop when it is necessary to facilitate this in data contracts and logical models. What is your opinion on this? What should we call this? Is it appropriate for their organization or can it be shared with others? They are considered to be shared and they are considered to be able to answer how a service operator works and who is involved in this service. Very good. We talked a little about the strategic approach that they have taken in cloud-first approach. Is this a pre-setting to get to the standardization that they want? It is not a pre-setting. We are in the national management system but I would actually say that there is a possibility to use machine tools and all the modern applications that are not traditionally on-prem gives us a lot of new possibilities. In the past, we have registered jobs with both data and development in a local environment on production data on sample data while we are now moving all the work that is done in a smart environment with all the nice tools for automated code and security patching so that we can get such services. We follow the national strategy and at the same time the first step is a bit of the strategy that we are going to be as efficient as possible and that it is going to be as long as it is done, so I am going to use the first step. So you know this well. And then we are going to use the field components. And then in addition to that this is the most important setting that we use tools that are also important in the strategy. So we use tools we are responsible for local development tools in this process that support the methodology we have chosen to make both tablets and on-board new people in need, perhaps the most useful in those who are the strongest on the methodology because the methodology is to support us so that we do not fail in the work again. So it is an important part of it when we go further with this. It is very exciting to see when you talk about a standardization of infrastructure using field components and standardization of tools we are actually in the middle of the topic we are talking about. There are many factors of work that you have had to do with standardization and there are some that come from the national management but some have also been intermed. But there is a very complex question but Jürgen, why is standardization important? Especially because there are so many actors that we have to get data from so it is necessary to come together and what we have chosen to focus on are experiences for others so we have seen what has been chosen to be done and well documented methodically and in the environment so we can learn and see how others have solved it and if we encounter a problem then it seems like someone has encountered a problem and we have solved it and in Ufiglösa B there is a smart follow-up and they are there for a reason we release the thought on the potential problems that can arise if we just follow the method because it is thought again and again with others before us therefore, as Martin talked about the long-code equipment we now have a way that standardizes the way what is called the logical model is the data contract for incentives which is a structured format on what they can and what cannot be seen in us and as we have a strong focus on the fact that there should be global facilities ready so we can get new incentives to the data but also identify if we have the same data several times and can measure it except for the low-code equipment it generates so automatically I would like to say that when we have thought about a structure that is related to a certain area or a certain feeling we have generated a code that publishes it and makes it clear to modulate the effective manufacturing in the register and at the same time we must take care of it and the contract towards the external actors so that is both the standardization and the establishment of good methods and a best practice and then we talked about the work process and how they do it and how they have introduced some agile work processes and then the standardization can you tell us a little more about it Jörgen? Yes We have created the logical model and we need the data that we need to answer the user stories that we have discussed and then we have another work for modeling that we have chosen for the effective manufacturing and the methodology and the data was 200 and there we also have a standardization a new logical model, a new shield and how it can be made effectively The work that I follow is a method that is quite strict so we can not mess with it and the work that I do also generates all the code for processing and orchestrating of the code furthermore it is an effective process that we can also generate cymbal models dimension models or whatever it is so we can also evaluate the consistency and show that we can not get it based on the structure we have set and that we can quickly correct before we proceed and begin to ask the data just that the step there is quite important to go through because if you just look at the database it is a bit difficult to think about how to use it but when you first start to think about what you can group on what you can filter on how to arrange in different ways then you may see that you are doing something small on the management of the data to be able to answer the questions and we do that quickly and hopefully we will solve in the next model furthermore it is important that in the future there will be a very scaleable work process but if we have set it or corrected it then it will help us to change it and do it effectively it is super exciting and there are many things we can talk more about I really want to come back to the data vault a bit later in the meeting but before that the logic that you are picking up and it can secure it in the modern market here it is absolutely useful to talk with we have rectified the flag-referencing group we have talked with the health service so we have started with the different areas and three areas that we are trying to take into account here and we have rectified the flag-referencing group and we have also talked to work with other national actors for standardization it is very important to get this together and there are more places that need to be done to get better very exciting and I also like as the introduction I talked about the focus on data modeling and in earlier episodes maybe to challenge you a little bit I had a very good conversation about what data modeling is actually about we are able to secure the needs we have in a different way than to model we model just for people not for people, of course or we have modeling and there was a very good conversation where we talked about how the way for the general logic the general logic is for example vector data bases I am not completely sure about this I see both in modeling and in terms of understanding for people, we are also using machines do you have any thoughts about this? maybe it is a bit of a classic about data modeling the first thing I think is data quality is that it is actually a combination that we can say something about because it has been what is data quality actually but then you work so well with understanding of the relationship with them at the start and that it follows through the whole course but it is not the end where the data is and how they have gone through in terms of actually I hear Heine and is from the work process and that is what I have defined out in the healthcare industry that is the first thing that can be done it is something that is exciting that is often in the discussions you have around data catalog for example, you miss it you talk a lot about that data catalog can help to establish a data lineage over time or over the applications but there is something else that is up-to-date on data provenance which is very close to what you talked about earlier when you talked about reference groups and the management logic that you are able to pick up and it is very important to understand what it is that is actually what we are working on absolutely, absolutely correct and it is an important reason that you can actually use the data well and answer it as we talked about in the beginning we use the history and answer it so that it is connected and also that we use the data catalog in the status of the healthcare industry we use it already and we define the concept and the relationship between them and not, as it is usually done when you have a data catalog but then you don't have the questions about what has happened in the past so we use the data catalog at the start of the process for the data very good I am sorry, I have come back to Data World and not least because I think this is very exciting maybe also because I feel that Data World is not as widespread in the north or Scandinavia than in other places in the world where does Data World use their data and what advantages and disadvantages have you seen? Yes, we have a challenge with the fact that we have so many different areas of data from and very, very many that you have, several thousand that is one of the advantages of the Data World methodology if you have very many different fields very many different structures that you should gather in a register then the methodology is very good if you have a register or a Data World that is constant under change then the methodology is very good it is a great format to change except that you have to register in a good way the important aspect for us is to have an effective understanding of the technology and that is supported by Data World so a lot of parallel processing adapts and makes it possible and is really a short summary of what can be done for Data World and then you have something behind with Data World that makes you need to work on the top end because there are so many differences between the two and it is incredibly difficult to secure the code that you made following the methodology since then there are so many different tables and load processors but with the top end that generates the code it is very good that was probably a very important point that you need something that actually supports that you get the insight that you are out there trying to get the answer to the top end but then at the end we do a little practice that I would like to do at the end of the episodes and it is often to come with a code to action or a summary from the top end maybe you Marta can start and tell us what you think is important people to take with you Yes, I think it is important to have a working strategy that is to have a strategy for how you need it to be able to go a bit to the field and to have a solution and then you have a completely additional approach to how you should work with it and that it is about infrastructure and the technical platform but absolutely the most important is this conceptual modelling and describe the understanding and the context between them and that we always think of the work that comes after a little self-defense Very good Do you have anything to say, Jörgen? Yes I think it is important to take into account that you have a good understanding of the difference and a understanding of the data you need to gather and take some time before you establish a data platform but actually we need to take some time to be more effective and meet the needs and the answers you want to report on Very good Thank you very much for the meeting Thank you for inviting me