A million dollar project aims to expose bad medical research

A new initiative aims to monitor monitoring behind the retreat to defective or fake medical science research to nearly one million dollars.

the Scientific Integrity Center I just launched the medical evidence project, a two-year effort to determine the published medical research with a negative impact on health guidelines-and make sure that people are already hearing it.

Provided with a $ 900,000 grant of open charitable works and a key team that includes up to five investigators. The project will use forensic medicine tools to determine problems in scientific articles, and report the results reached by monitoring the decline, which is the first location of scientific monitoring.

“We originally established the Center for Scientific Integrity as a home to the hour of retreat mail Announcing the grant. “The medical evidence project allows us to support cash analysis and publish results.”

according to natureThese defective and confusing documents are confusing because they distort the adolescent-references that combine results from multiple studies to extract strong statistical conclusions. If two or two studies are made of two brackets in metaphorical analysis, they can direct standards for health policy.

In 2009, for example, the European guideline recommended the use of beta blockers during non -heart surgery, based on research at the beginning of the millennium that was questioned later. Years later, an independent review indicated that the guidance may have contributed 10,000 deaths annually in the United Kingdom.

Under the leadership of James Heathrez, the Science Safety Adviser, the team’s plan is to build software tools, and the expected customers of the unknown informants, and pushed peer auditors to verify their work. They aim to determine at least 10 defective twinous analyzes per year.

The team chooses its moments wisely. Gizmodo previously mentioned, Unwanted science from artificial intelligence The academic digital ecosystem is immersed, and in everything from the conference’s procedures appears to magazines that the pendant reviews. A study published in the review of the wrong information of Harvard Kennedy School found that two-thirds of the papers from which samples were taken through the scientific researcher from Google contain signs of the text created by GPT-even in the prevailing scientific outlets. About 14.5 % of these false studies focused on health.

This is particularly concerned because Google’s scientific researcher does not distinguish between studies reviewed by peers with prior missions, student papers or other active works. Once this type of hunting is withdrawn to a metaphysical analysis or cited doctors, it is difficult to decipher the consequences. “If we cannot trust that the research we read is real,” one of the researchers told Gizmodo, “We are risking decision -making based on incorrect information.”

We have already seen how nonsense can escape. In 2021, nature Springer More than 40 declined Papers from the Arab Sciences for Geographic Sciences-very improper studies that they read, such as crazy LIBS, which was created from artificial intelligence. Only last year, the publisher had to withdraw a paper that includes impossible images of artificial intelligence. Genital mice.

We have introduced the era of digital excavations, where artificial intelligence models began Preserving and spreading Nonsent phrases as if they were real scientific terms. For example, earlier this year, a group of researchers found a set of distorted words of the biology paper in 1959 included in the outputs of large language models including the GPT-4O of Openai.

In this climate, the goal of the medical evidence project appears to be sorting more than cleaning. The team deals with a set of defective information, hiding in sight, and many of them can have very real health consequences if taken by the nominal value.