Health data can greatly benefit the health sector in the form of digital tools. But what exactly is health data, and how can it be utilised?
– Data driven solutions are used in more and more industries, and finally the healthcare sector has started to see the advantages as well, says Håkon Lorentzen, CTO at Deepinsight.
The benefits of a more data-driven health sector are, amongst other things, an increased insight into medical connections, more efficient workdays for healthcare professionals, better use of resources and, most importantly, better patient care.
– “Health data” is a broad term. One could assume that health data only includes what’s written in different health registries about the patient’s medical history. But in reality, health data is a lot more, says Håvard Thøgersen, COO at Deepinsight.
Administrative data – drift og planlegging
We often split health data into two main categories: clinical data and administrative data. In addition, registry data is an important subcategory.
Administrative data explains the patient’s journey through for example a hospital. This includes data showing when the patient was at the hospital, which departments the patient was in contact with, who the patient met with and what was done. The data covers both the planning in advance, and what actually happened. Such data is found in different digital systems, for example DIPS.
With access to administrative data, Deepinsight can give employees a basis for decision making regarding the planning of patient care. That would contribute to a more sensible use of resources in the health sector.
Clinical data – patient information
Clinical data describes the patient’s medical history and healthcare professionals’ assessments. Such data is found in patient journals, general physician’s digital systems, home monitoring systems, test results, lab rapports and more. Lab reports also contain valuable genetic data.
Luckily, most of the clinical data is stored digitally. The challenge with using the data is poor structure. Different terminologies are used, a lot of it is subjective and there are great amounts of freestyle text.
A good example is the patient journal system. Here, you can find a great amount of valuable data, but today the system is used more as a communication tool, where healthcare professionals write notes to each other.
There are also different specialised systems that store data manually, where algorithms have to be put to use, such as for X-rays.
– Our most important task is to systematize existing data in order to create high quality datasets. To create order in the caos, literally, explains Severin Sjømark, Chief Data Officer at Deepinsight.
Registry data – structured clinical data
Registry data is mainly data from the many health registries we have. In addition to national health registries and medical quality registries, there are also several smaller quality registries, bio banks and health examinations that can be very useful.
Examples of health registries are the Norwegian Patient Registry, the Cancer Registry Norway and the Norwegian Prescription Database.
Data gathered from health registries is usually structured and detailed. The registries are widely used for science, especially in order to increase the understanding of correlations, or epidemiology. This will raise the quality of the diagnostics and treatments of diseases.
What can we use health data for?
When health data is systematized and structured, it increases the potential of developing tools based on machine learning that can contribute to automate work processes in the healthcare sector. This includes matching patients with resources, filling out journals and picture processing. Such tools will lead to more efficient operations and faster diagnosis.
Quality data can both be built and analysed with machine learning and thereafter be used for research and clinical studies. This method leads to research of high quality and reduces the development time of the datasets significantly.
– Today, a PhD candidate uses great parts of the PhD to read through and structure data for research purposes. Machine learning can change that, says CTO Lorentzen.
How to access health data
For Deepinsight to be able to succeed with the development of health tools based on data, direct access to different databases with journals, patient data and administrative data is necessary. There are strict legal and technical requirements for sharing and using such health data, and it is the health institutions, as the owners of the data, who are responsible for making sure the data sharing happens in a secure manner.
– I think there’s still a way to go before data-driven tools are standard in the healthcare sector. The combination of time consuming processes and strict rules around the use of health data makes it complicated, says Thøgersen and continues:
– We have a long-term perspective. A lot will happen in the next decade. The patient journey and the workday of healthcare professionals will be completely different. Deepinsight will be a big part of that.