Strategy proposal for further implementation of machine learning in the Cluster of Reviews and Health Technology Assessments
We have since 2020 implemented machine learning (ML) in the production of evidence syntheses and health technology assessments. This was due to a need and desire to streamline such research processes, as the gold standard methods are resource-intensive, making current practices unsustainable.
Since 2020 the Cluster for Reviews and Health Technology Assessments (HTV) has implemented machine learning (ML) in the production of evidence syntheses and health technology assessments. This was due to a need and desire to streamline such research processes, as the gold standard methods are resource-intensive, making current practices unsustainable. ML can automate complex, repetitive tasks in evidence synthesis processes, thus reducing resource requirements.
HTV established two dedicated teams for the implementation of ML: ML 1.0 (2020-2021) and ML 2.0 (2021-2022). The ML teams have been very successful with this work: they have documented workload savings, as well as established themselves as implementation leaders in the field. The experiences from the ML work in HTV since 2020 form the basis for this report, which describes ML 2.0's strategy proposal for how HTV should implement further work with ML in evidence synthesis processes:
- ML 3.0 maintain focus on exploration, to identify new functions and applications.
- Existing HTV expertise structures (e.g. undervisningslaget) take responsibility for building employee capacity to use established functions.
- Team 3.0 retains responsibility for building capacity of novel functions and applications, and all such training must be scalable, e.g. asynchronous and interactive online trainings.
A final suggestion is for HTV to collect existing and new development and innovation activities into one portfolio in the cluster. This portfolio could include ML Team 3.0, as well as other types of projects, teams, and activities related to automation, digitalization, and process change. An area of co-generative learning would be created, as well as an incubator for funding applications.