Claims of faked science are not fake news - researchers use AI, admit to falsifying data

Everybody gains from the publication game apart from the patients who suffer from being given treatments based on fraudulent data.

Ian Roberts, professor of epidemiology at the London School of Hygiene & Tropical Medicine as quoted by Richard Smith, former BMJ editor 

Artificial Intelligence corrupts medical publishing

Wiley science publisher has recently retracted 11,000 “peer-reviewed” papers found to be fake, causing the longtime publisher to shut down 19 of its journals. Quoting Australian blogger Jo Nova about the scandal, ZeroHedge reported:

217 year old Wiley science publisher has reportedly "peer reviewed" more than 11,000 papers that were determined to be fake without ever noticing. The papers were referred to as "naked gobbledygook sandwiches",  Australian blogger Jo Nova wrote on her blog last week.

According to Nova, as author Tyler Durden reported, these papers were being written by artificial intelligence and, seemingly, did not go through “peer-review” by human beings:

According to Nova's post, professional cheating services are employing AI to craft seemingly "original" academic papers by shuffling around words. For instance, terms like "breast cancer" morphed into "bosom peril," and a "naïve Bayes" classifier turns into "gullible Bayes."

Similarly, in one paper, an ant colony was bizarrely rebranded as an "underground creepy crawly state."

The misuse of terminology extends to machine learning, where a 'random forest' is whimsically translated to 'irregular backwoods' or 'arbitrary timberland'.

Nova writes that shockingly, these papers undergo peer review without any rigorous human oversight, allowing egregious errors, like converting 'local average energy' to 'territorial normal vitality', to slip through.

The following text and image were part of an article published by ARSTechnica showing AI-generated figures in a peer-reviewed article. Scientists were appalled by the image that the authors admitted was made by Midjourney, a generative artificial intelligence program.

Inherently flawed research not new

Many medical journals publish inherently flawed research that is harder to detect and have been doing so for many, many years. Frontline News addressed this issue at the time that “data sleuth” Sholto David “called out four prominent research scientists with Harvard Medical School's Dana-Farber Cancer Institute (DFCI) for ‘data forgery.’”

Authorship fraud

Frontline News noted that a “publish or perish” culture in academia often leads to fraud since professors and scientists must publish to advance their careers and to receive grant funding. One type of fraud perpetrated is authorship fraud, where instructors and professors "who may not even be aware of the work being done” take credit for papers that were written by their students, who remain anonymous. 

Another type of authorship fraud is perpetrated by pharmaceutical companies employing medical ghostwriters. Pharma pays individuals to write articles in favor of their products and pays doctors to attach their names as authors. 

Medical ghostwriters are employed by pharmaceutical companies and medical-device manufacturers to produce apparently independent manuscripts for peer-reviewed journals, conference presentations and other communications. Physicians and other scientists are paid to attach their names to the manuscripts as though they had authored them. The named authors may have had little or no involvement in the research or writing process.

Editors of medical journals and government agencies are well aware that a lot of what's published cannot be trusted. Frontline News reported on the “confession” of Richard Horton, editor-in-chief of the medical journal Lancet, who stated, in an April 2015 article, that probably half of what's published in the medical journals is simply not true. 

The case against science is straightforward: much of the scientific literature, perhaps half, may simply be untrue. Afflicted by studies with small sample sizes, tiny effects, invalid exploratory analyses, and flagrant conflicts of interest . . . science has taken a turn towards darkness. . . .  scientists too often sculpt data to fit their preferred theory of the world. Or they retrofit hypotheses to fit their data. Journal editors . . . . love of “significance” pollutes the literature with many a statistical fairy-tale. We reject important confirmations. . . 

Not possible to believe much of published research

Dr. Marcia Angell talked about the problem of relying on published papers. The former editor-in-chief of the NEJM (New England Journal of Medicine stated in 2002:

It is simply no longer possible to believe much of the clinical research that is published, or to rely on the judgment of trusted physicians or authoritative medical guidelines. I take no pleasure in this conclusion, which I reached slowly and reluctantly over my two decades as an editor

FDA big part of the problem

Frontline News wrote about the FDA's failure to notify the public and journals when it uncovers fraud in clinical trials, referencing a 2015 paper in the BMJ which explained that it was not solely due to lack of oversight of clinical trials but also resulted from the FDA's “keeping scientific misconduct from the medical establishment.”

The great extent of the FDA's treachery was written up in a separate 2015 paper published by JAMA (Journal of the American Medical Association) and described in an article for Slate Magazine by the paper's author, Charles Siefe. It is so bad that not only doesn't the FDA tell its advisory group, which is deciding on allowing drugs to be marketed, about problems with drug trials, but it even refused to tell Congress what it found out about problems with the clinical trials of an antibiotic named Ketek:

About a decade ago, the agency got into trouble over a newly approved antibiotic, Ketek. Inspectors had found extensive problems (including fraud) affecting key clinical trials of the drug. Yet the agency did its best to hide the problems from even its most trusted advisers. 

As David Ross, the FDA official in charge of reviewing Ketek’s safety, put it, “In January 2003, over reviewers’ protests, FDA managers hid the evidence of fraud and misconduct from the advisory committee, which was fooled into voting for approval.” However, when the reports of misconduct at one clinical site began appearing in the press—along with stories of liver damage and blurred vision associated with the new drug—Congress stepped in, demanding information from the agency about the fraud.

But even the Senate couldn’t wring key information about the misconduct out of the FDA. “Every excuse under the sun has been used to create roadblocks,” complained an indignant Sen. Charles Grassley, “even in the face of congressional subpoenas requesting information and access to FDA employees.” The head of the FDA, Andrew von Eschenbach, attempted to explain to Congress why the agency didn’t tell its advisory committee about the problems in the Ketek study: “After considering the fact that the investigation results were preliminary … FDA decided to hold the Advisory Committee meeting as planned …” without notifying the committee of the potential problems. But Rep. Bart Stupak quickly pointed to an email, which, he argued, contradicted von Eschenbach’s testimony. “So either you are not being forthright with us, when I believe you are, but whoever is doing your work is trying to  lead this committee down the wrong path.” And the correct path showed that site after site involved in study 3014, as well as other key Ketek studies, were tainted as well. (Emhasis added.)

“Researchers admitted to falsifying data”

A Columbia University paper also detailed the extent of the problem regarding research, which seems intractable considering the scope of the problem and how endemic it seems to be:

. . .  8 percent of scientists and 10 percent of medical and life-sciences researchers admitted to falsifying data at least once between 2017 and 2021 in a Dutch study of 6,813 researchers, while more than half engaged in at least one questionable research practice. Questionable research practices include research design flaws or unfairness in decisions surrounding publication or grants.. . .  3,000 scientists with NIH grants in the United States, 0.3 percent of the scientists responding admitted fabricating research data and 1.4 percent of them admitted plagiarizing.

. . .  many scientists admitted that they engaged in a range of behaviors beyond fabrication, falsification, and plagiarism that undermine the integrity of science, such as changing the results of a study under pressure from a funding source or failing to present data that contradicts one’s previous research. 

Frontline News also reported a bit of good news in the quest for reliable published research. 

Elsevier is working to eliminate fraud in publishing. Elsevier publishes 2,800 journals and employs a Research Integrity and Ethics team that is moving to a proactive stance to identify fraudulent papers before they are published.

Journal editors culpable, too

The problem may not be solvable as long as the FDA remains captured by Pharma and the editors of medical journals receive huge sums from Pharma as well, as reported in 2018 by Science Alert:

Here's just a small highlight showing editor payments received in 2014 - the amount on the left is direct payments, and the 'research' payments on the right are less regulated, usually made in the form of expensive research trips. 

Fraud takes on life of its own

Even when fraud is detected and the papers are retracted, fraudulent papers have often already developed a life of their own, having been cited in other papers, used as the basis for other research, and having found their way into medical decision-making, as reflected by this article's introductory quote by Professor Roberts. 

How much fraud is never detected?