The German Federal Ministry of Health's (G-BA) and Laser AI
Based on a presentation given at IRM 2022: “I never thought about that before”: Insights from a development process for a screening tool
Background
The German Federal Ministry of Health's (G-BA) team comprises 25+ scientists. They are responsible for processing over 200 clinical questions each year following the principles of evidence-based medicine. In 2019, the team sought a tool to accelerate their literature selection process based on title and abstract screening. Their search led to an extensive evaluation process, ultimately resulting in Laser AI's selection.
The G-BA team initially evaluated 43 software applications, shortlisting eight after preliminary testing. Two of these options were selected for final testing. After the final comparison, Laser AI was chosen due to its easy data-importing capabilities and convenience in screening.
Adoption
The user-centred design methodology was applied during the development process. It allowed for close cooperation with the G-BA team to build features that aligned well with their needs. The Evidence Prime staff conducted requirement analysis during the implementation process to map out the specifics of the G-BA’s workflow. The tool was then configured to fulfil the requirements regarding import and export formats and user permissions. For example, one of the addressed needs was the possibility for the management staff to see the list of all projects within the organization to be able to intervene and take over if the original project lead is temporarily unavailable. Additionally, the user interface and notification emails were translated into German.
After a nine-month rollout, the G-BA team evaluated Laser AI. The results were overwhelmingly positive with:
89% of scientists confirming that the quality of literature selection had improved with the use of the Laser AI software. Additionally, 80% of scientists found Laser AI more manageable and flexible than the previous proprietary solution.
The researchers could leverage machine learning in their screening process for the first time, thanks to the Laser AI’s built-in reference ranking model. Project managers valued the tool’s flexibility in handling situations such as changes in the project team, allowing them to either retain or erase screening decisions made by members leaving the project.
These findings are a testament to Laser AI's effectiveness in streamlining the literature selection process. Laser AI's compliance with government organisations' most stringent standards made it a perfect fit for the G-BA team. The tool's ability to quickly adapt to existing review processes was also a significant factor in its selection.
Conclusion
The German Federal Ministry of Health's decision on software was based on a rigorous evaluation process that involved testing and comparing several software applications.
Laser AI's ease of data importing, convenience in screening, and compliance with government standards made it the clear choice for the G-BA team. The user-centred design methodology applied during the development process ensured that the final tool met the users' needs and was easy to use.
The positive evaluation results confirmed that Laser AI had significantly improved the quality of literature selection and was more user-friendly and flexible than the previous proprietary solution.
Laser AI is a valuable tool for organisations looking to streamline their literature selection process, with its ability to easily adapt to existing review processes and its compliance with government standards.