Scientists develop huge memory for developing effective policies

Scientists develop huge memory for developing effective policies

"In the end, we could benefit enormously from a world in which comprehensive evidence syntheses are available in one place that is continuously updated about every important social problem," says Will Moy, who heads Campbell Collaboration, an international non -profit organization that supports social science reviews.

Although researchers create a large number of studies in political areas, syntheses that represent the weight of evidence on a topic are rare in many areas and are not routinely used for the guideline for political design. "There is enormous demand" by political decision -makers for such syntheses, Jen Gold, director of research at the Economic and Social Research Council (ESRC), says a British funding agency. "But the offer does not coincide."

evidence syntheses are "everything that the world knows about how to solve an important problem in one place," explains Moy. In medicine, doctors routinely use thousands of systematic overviews- Thorough Synthesis - that show whether treatment helps or harms. In most other areas, however, there is a lack of such an extensive basis (see 'missing syntheses'). It can take months or years to extract importance from a huge research corpus - and the funding providers have historically issued a comparatively little for the synthesis of knowledge, compared to the billions that they spend on new research.

In order to counteract this, the ESRC and Wellcome, the biomedical research promoter in London, announced on September 21 that they invest 9.2 million GBP (12.2 million USD) or around 45 million GBP in databases and tools for over five years, which can help with the synthesis of research. The British Minister of Science Patrick Vallance and well come boss John-Arne Røttingen announced the financing at an event in New York, which was associated with the summit of the United Nations to the future, a meeting that aims to design a better world, including science.

researchers have welcomed the news - it is assumed that this is one of the greatest individual investments in evidence synthesis - and say that it comes at the right time because progress accelerates the process of finding and combining studies. "It's so exciting," says Isabelle Mercier, a researcher at the United Nations development program, the Evidence Syntheses supported . "It was too big four years ago to think about it, but now we start to see that we can actually implement it," she says.

But AI also makes the task more difficult because AI chatbots such as chattgpt can create credible-sounding but potentially misleading summaries of research results. "The challenge is how to take off what is really reliable, what it is not," says Moy.

The production of syntheses is usually slow, difficult and expensive. Researchers who carry out a systematic overview must search for databases worldwide for published and unpublished work in order to find potentially relevant studies. Then reduce a long list of thousands of studies to the most relevant, evaluate their reliability, extract the data and combine the results, sometimes using a statistical method called meta analysis. Even when they are completed, evidence syntheses often do not reach political decision -makers and quickly observe while new research results arrive. "If a political decision -maker comes with a question, it shouldn't take three months to find research," says James Thomas, a specialist in research synthesis at the University College London. "That's ridiculous."

The problem tightened during the Covid 19 pandemic when the authorities needed fast syntheses everywhere to make decisions about medication, masks and lockdowns. At first the scientists could not provide them quickly enough-but then they produced Too many double syntheses and poor ratings .

The dream of the scientists is that everyone, everywhere, a synthesis that is tailored to their question and their region could put together a push of a button. For this purpose, the researchers would like to create 'Evidence Banks': common databases with pre -selected studies that are tagged with information such as method and location and contain data in the common format so that they can be combined. Trained AI tools should take over most of the tedious task of sorting studies and the synthesis of data, while people check quality-for example by evaluating possible distortions in the underlying studies.

Some databases are already on the right track. The Education Endowment Foundation (EEF), a charity in London, has a database with more than 3,500 education studies. With the help of this database, the organization has created a number of systematic overviews that the reveal Influence of tutoring, homework and class size on learning . It shares the database and overviews with several countries to avoid repeating others. Ideally, "instead of creating six separate systematic overviews, you create a great overview that we can use together," says Jonathan Kay, who heads the work to summarize evidence at EEF.

The latest investments could finally lead to a number of databases that are similar to the EEF and are willing to be synthesized for important policy areas such as environmental protection. Out of these, supporters want to build "living"-or constantly updated-evidence syntheses that show, for example, What improves mental health and reduces youth unemployment.