Researchers develop a machine learning framework to help with research

Researchers at Utrecht University have recently developed a machine learning framework, called ASReview, that can automatically run through numerous past studies and compile high-quality literature reviews. This could be a game-changer for researchers and scientists, especially for those working on the pandemic.

Indeed, the framework, developed by Jonathan de Bruin and a team of researchers, will be able to automatically recognize the most relevant studies about a topic by training the machine learning model with an interactive approach called active learning. Hence, this system is trained to analyze various documents available and find the relevant ones.

In times of a pandemic, new guidelines and searches were required so the researchers worked with the Allen Institute for AI to develop the framework. ASReview has then been used by many researchers to review past studies about the virus and find more efficient medical guidelines. It is predicted that in the future, it will conduct several systematic reviews and meta-analyses to speed up research in various fields.

The researchers do believe that interactive machine learning approaches have to remain completely transparent and explainable.