Tag: Publishing

China has a list of suspect journals and it’s just been updated

[ad_1]

The National Science Library of the Chinese Academy of Sciences in Beijing.Credit: Yang Qing/Imago via Alamy

China has updated its list of journals that are deemed to be untrustworthy, predatory or not serving the Chinese research community’s interests. Called the Early Warning Journal List, the latest edition, published last month, includes 24 journals from about a dozen publishers. For the first time, it flags journals that exhibit misconduct called citation manipulation, in which authors try to inflate their citation counts.

Yang Liying studies scholarly literature at the National Science Library, Chinese Academy of Sciences, in Beijing. She leads a team of about 20 researchers who produce the annual list, which was launched in 2020 and relies on insights from the global research community and analysis of bibliometric data.

The list is becoming increasingly influential. It is referenced in notices sent out by Chinese ministries to address academic misconduct, and is widely shared on institutional websites across the country. Journals included in the list typically see submissions from Chinese authors drop. This is the first year the team has revised its method for developing the list; Yang speaks to Nature about the process, and what has changed.

How do you go about creating the list every year?

We start by collecting feedback from Chinese researchers and administrators, and we follow global discussions on new forms of misconduct to determine the problems to focus on. In January, we analyse raw data from the science-citation database Web of Science, provided by the publishing-analytics firm Clarivate, based in London, and prepare a preliminary list of journals. We share this with relevant publishers, and explain why their journals could end up on the list.

Sometimes publishers give us feedback and make a case against including their journal. If their response is reasonable, we will remove it. We appreciate suggestions to improve our work. We never see the journal list as a perfect one. This year, discussions with publishers cut the list from around 50 journals down to 24.

Yang Liying studies scholarly literature at the National Science Library and manages a team of 20 to put together the Early Warning Journal List.Credit: Yang Liying

What changes did you make this year?

In previous years, journals were categorized as being high, medium or low risk. This year, we didn’t report risk levels because we removed the low risk category, and we also realized that Chinese researchers ignore the risk categories and simply avoid journals on the list altogether. Instead, we provided an explanation of why the journal is on the list.

In previous years, we included journals with publication numbers that increased very rapidly. For example, if a journal published 1,000 articles one year and then 5,000 the next year, our initial logic was that it would be hard for these journals to maintain their quality-control procedures. We have removed this criterion this year. The shift towards open access has meant that it is possible for journals to receive a large number of manuscripts, and therefore rapidly increase their article numbers. We don’t want to disturb this natural process decided by the market.

You also introduced journals with abnormal patterns of citation. Why?

We noticed that there has been a lot of discussion on the subject among researchers around the world. It’s hard for us to say whether the problem comes from the journals or from the authors themselves. Sometimes groups of authors agree to this citation manipulation mutually, or they use paper mills, which produce fake research papers. We identify these journals by looking for trends in citation data provided by Clarivate — for example, journals in which manuscript references are highly skewed to one journal issue or articles authored by a few researchers. Next year, we plan to investigate new forms of citation manipulation.

Our work seems to have an impact on publishers. Many publishers have thanked us for alerting them to the issues in their journals, and some have initiated their own investigations. One example from this year, is the open-access publisher MDPI, based in Basel, Switzerland, whom we informed that four of its journals would be included in our list because of citation manipulation. Perhaps it is unrelated, but on 13 February, MDPI sent out a notice that it was looking into potential reviewer misconduct involving unethical citation practices in 23 of its journals.

You also flag journals that publish a high proportion of papers from Chinese researchers. Why is this a concern?

This is not a criterion we use on its own. These journals publish — sometimes almost exclusively — articles by Chinese researchers, charge unreasonably high article processing fees and have a low citation impact. From a Chinese perspective, this is a concern because we are a developing country and want to make good use of our research funding to publish our work in truly international journals to contribute to global science. If scientists publish in journals where almost all the manuscripts come from Chinese researchers, our administrators will suggest that instead the work should be submitted to a local journal. That way, Chinese researchers can read it and learn from it quickly and don’t need to pay so much to publish it. This is a challenge that the Chinese research community has been confronting in recent years.

How do you determine whether a journal has a paper-mill problem?

My team collects information posted on social media as well as websites such as PubPeer, where users discuss published articles, and the research-integrity blog For Better Science. We currently don’t do the image or text checks ourselves, but we might start to do so later.

My team has also created an online database of questionable articles called Amend, which researchers can access. We collect information on article retractions, notices of concern, corrections and articles that have been flagged on social media.

Source: Early Warning Journal List

What impact has the list had on research in China?

This list has benefited the Chinese research community. Most Chinese research institutes and universities reference our list, but they can also develop their own versions. Every year, we receive criticisms from some researchers for including journals that they publish in. But we also receive a lot of support from those who agree that the journals included on the list are of low quality, which hurts the Chinese research ecosystem.

There have been a lot of retractions from China in journals on our list. And once a journal makes it on to the list, submissions from Chinese researchers typically drop (see ‘Marked down’). This explains why many journals on our list are excluded the following year — this is not a cumulative list.

This interview has been edited for length and clarity.

[ad_2]

Source link

March 6, 2024
Nature publishes too few papers from women researchers — that must change

[ad_1]

Women and early-career researchers: Nature wants to publish your research.Credit: Getty

Researchers submitting original research to Nature over the past year will have noticed an extra question, asking them to self-report their gender. Today, as part of our commitment to helping to make science more equitable, we are publishing in this editorial a preliminary analysis of the resulting data, from almost 5,000 papers submitted to this journal over a five-month period. As well as showing the gender split in submissions, we also reveal, for the first time, possible interactions between the gender of the corresponding author and a paper’s chance of publication.

The data make for sobering reading. One stark finding is how few women are submitting research to Nature as corresponding authors. Corresponding authors are the researchers who take responsibility for a manuscript during the publication process. In many fields, this role is undertaken by some of the most experienced members of the team.

The giant plan to track diversity in research journals

During the period analysed, some 10% of corresponding authors preferred not to disclose their gender. Of the remainder, just 17% identified as women — barely an increase on the 16% we found in 2018, albeit using a less precise methodology. By comparison, women made up 31.7% of all researchers globally in 2021, according to figures from the United Nations science, education and cultural organization UNESCO (see go.nature.com/3wgdasb).

Large geographical differences were also laid bare. Women made up just 4% of corresponding authors of known gender from Japanese institutions. Of researchers from the two countries submitting the most papers, China and the United States, women made up 11% and 22%, respectively. These figures reflect the fact that women’s representation in research drops at the most senior levels. They also mirror available data from other journals¹, although it is hard to find direct comparisons for a multidisciplinary journal such as Nature.

At Cell, which has a life-sciences focus, women submitted 17% of manuscripts between 2017 and 2021, according to an analysis of almost 13,000 submissions². The most recent data on gender from the American Association for the Advancement of Science (AAAS), which publishes the six journals in the Science family, is collected and reported differently. Some 27% of their authors of primary and commissioned content, and their reviewers, are women, according to the AAAS Inclusive Excellence Report (see go.nature.com/3t6yyr8). Nonetheless, all of these figures are just too low.

Another area of concern is acceptance rates. Of the submissions included in the current Nature analysis, those with women as the corresponding author were accepted for publication at a slightly lower rate than were those authored by men. Some 8% of women’s papers were accepted (58 out of 726 submissions) compared with 9% of men’s papers (320 out of 3,522 submissions). The acceptance rate for people self-reporting as non-binary or gender diverse seemed to be lower, at 3%, although this is a preliminary figure and we have reason to suspect that the real figure could be higher, as described below. Once we have a larger sample, we plan to test whether the differences are statistically significant.

Sources of imbalance

So, at what stage in the publishing process is this imbalance introduced? Men and women seem to be treated equally when papers are selected for review. The journal’s editors — a group containing slightly more women than men — were just as likely to send papers out for peer review for women corresponding authors as they were for men. For both groups, 17% of submitted papers went for peer review.

Our efforts to diversify Nature’s journalism are progressing, but work remains

A difference arose after that. Of those papers sent for review, 46% of papers with women as corresponding authors were accepted for publication (58 of 125) compared with 55% (320 of 586) of papers authored by men. The acceptance rate for non-binary and gender-diverse authors was higher at 67%. However, this is from a total of only three reviewed papers, a figure that is too small to be meaningful.

This difference in acceptance rates during review tallies with the findings of a much larger 2018 study of 25 Nature-family journals, which used a name-matching algorithm, rather than self-reported data³. Looking at 17,167 papers sent for review over a 2-year period, the authors found a smaller but significant difference in acceptance rates, with 43% for papers with a woman as corresponding author, compared with 45% for a man. However, they were unable to say whether the difference was attributable to reviewer bias or variations in manuscript quality.

Peering into peer review

How much bias exists in the peer-review process is difficult to study and has long been the subject of debate. A 2021 study in Science Advances that looked at 1.7 million authors across 145 journals between 2010 and 2016 found that, overall, the peer-review and editorial processes did not penalize manuscripts by women⁴. But that study analysed journals with lower citation rates than Nature, and its results contrast with those of previous work⁵, which found gender-based skews.

Moreover, other studies have shown that people rate men’s competence more highly than women’s when assessing identical job applications⁶; that there is a gender bias against women in citations; and that women are given less credit for their work than are men⁷. Taken together, this means we cannot assume that peer review is a gender-blind process. Most papers in our current study were not anonymized. We did not share how the authors self-reported, but editors or reviewers might have inferred gender from a corresponding author’s name. Nature has offered double-anonymized peer review for both authors and reviewers since 2015. Too few take it up for us to have been able to examine its impact in this analysis, but the larger study in 2018 looked at this in detail³.

Data limitations

There are important limitations to Nature’s data: we must emphasize again that they are preliminary. Moreover, they provide the gender of only one corresponding author per paper, not the gender distribution of a paper’s full author list. Furthermore, they don’t describe any other differences between authors.

There are also aspects of the data that need to be investigated further. For example, we need to look into the possibility that the option of reporting as non-binary or gender diverse is being misinterpreted by some authors with English as a second language. We think that ironing out such misunderstandings could result in a higher acceptance rate for non-binary authors.

Nature’s under-representation of women

Most importantly, these data give no insight into author experiences in relation to race, ethnicity and socio-economic status. Although men often have advantages compared with women, other protected characteristics also have a significant impact on scientists’ careers. Nature is participating in an effort by a raft of journal publishers to document and reduce bias in scholarly publishing by tracking a range of characteristics. This is a work in progress and sits alongside Springer Nature’s wider commitment to tackling inequity in research publishing.

So what can Nature do to ensure that more women and minority-gender scientists find a home for their research in our pages?

First, we want to encourage a more diverse pool of corresponding authors to submit. The fact that only 17% of submissions come from corresponding authors who identify as women might reflect existing imbalances in science (for example, it roughly tracks with the 18% of professor-level scientists in the European Union who are women, as reported by the European Commission⁸).

But there remains much scope for improvement. We know that the workplace climate in academia can push women out or see them overlooked for senior positions⁹. A 2023 study published in eLife found that women tend to be more self-critical of their own work than men are and that they are more frequently advised not to submit to the most prestigious journals¹⁰.

Second, just as prestigious universities should not simply lament their low application numbers from under-represented groups, we should not sit back and wait for change to come to us. To this end, our editors will actively seek out authors from these communities when at conferences and on laboratory visits. We will be more proactive in reaching out to women and early-career researchers to make sure they know that Nature wants to publish their research. We encourage authors with excellent research, at any level of seniority and at any institution, to submit their manuscripts.

Third, in an effort to make peer review fairer, Nature’s editors have been actively working to recruit a more diverse group of referees; 2017 data found that women made up just 16% of our reviewers. We need to double down on our efforts to improve this situation and update readers on our progress. In the future, we also plan to analyse whether corresponding authors’ gender affects the number of review cycles they face, and whether there are differences in relation to gender according to discipline and prestige of their affiliated institution. We need to improve our understanding of the sources of inequity before we can work on ways to address them. Nature’s editors will also strive to minimize our own biases through ongoing unconscious-bias training.

Last but not least, we will keep publishing our data on authorship and peer review, alongside complementary statistics on the gender of contributors to articles outside original research. Although today’s data present just a snapshot, Nature remains committed to tracking the gender of authors, to regularly updating the community on our efforts, and to exploring ways to make the publication process more equitable.

[ad_2]

Source link

March 6, 2024
More than 2 million research papers have disappeared from the Internet

[ad_1]

A study identified more than two million articles that did not appear in a major digital archive, despite having an active DOI.Credit: Anna Berkut/Alamy

More than one-quarter of scholarly articles are not being properly archived and preserved, a study of more than seven million digital publications suggests. The findings, published in the Journal of Librarianship and Scholarly Communication on 24 January¹, indicate that systems to preserve papers online have failed to keep pace with the growth of research output.

“Our entire epistemology of science and research relies on the chain of footnotes,” explains author Martin Eve, a researcher in literature, technology and publishing at Birkbeck, University of London. “If you can’t verify what someone else has said at some other point, you’re just trusting to blind faith for artefacts that you can no longer read yourself.”

Eve, who is also involved in research and development at digital-infrastructure organization Crossref, checked whether 7,438,037 works labelled with digital object identifiers (DOIs) are held in archives. DOIs — which consist of a string of numbers, letters and symbols — are unique fingerprints used to identify and link to specific publications, such as scholarly articles and official reports. Crossref is the largest DOI registration agency, allocating the identifiers to about 20,000 members, including publishers, museums and other institutions.

The sample of DOIs included in the study was made up of a random selection of up to 1,000 registered to each member organization. Twenty-eight percent of these works — more than two million articles — did not appear in a major digital archive, despite having an active DOI. Only 58% of the DOIs referenced works that had been stored in at least one archive. The other 14% were excluded from the study because they were published too recently, were not journal articles or did not have an identifiable source.

Preservation challenge

Eve notes that the study has limitations: namely that it tracked only articles with DOIs, and that it did not search every digital repository for articles (he did not check whether items with a DOI were stored in institutional repositories, for example).

Nevertheless, preservation specialists have welcomed the analysis. “It’s been hard to know the real extent of the digital preservation challenge faced by e-journals,” says William Kilbride, managing director of the Digital Preservation Coalition, headquartered in York, UK. The coalition publishes a handbook detailing good preservation practice.

“Many people have the blind assumption that if you have a DOI, it’s there forever,” says Mikael Laakso, who studies scholarly publishing at the Hanken School of Economics in Helsinki. “But that doesn’t mean that the link will always work.” In 2021, Laakso and his colleagues reported² that more than 170 open-access journals had disappeared from the Internet between 2000 and 2019.

Kate Wittenberg, managing director of the digital archiving service Portico in New York City, warns that small publishers are at higher risk of failing to preserve articles than are large ones. “It costs money to preserve content,” she says, adding that archiving involves infrastructure, technology and expertise that many smaller organizations do not have access to.

Eve’s study suggests some measures that could improve digital preservation, including stronger requirements at DOI registration agencies and better education and awareness of the issue among publishers and researchers.

“Everybody thinks of the immediate gains they might get from having a paper out somewhere, but we really should be thinking about the long-term sustainability of the research ecosystem,” Eve says. “After you’ve been dead for 100 years, are people going to be able to get access to the things you’ve worked on?”

[ad_2]

Source link

March 4, 2024
Is ChatGPT making scientists hyper-productive? The highs and lows of using AI

[ad_1]

In a 2023 Nature survey of scientists, 30% of respondents had used generative AI tools to help write manuscripts.Credit: Nicolas Maeterlinck/Belga MAG/AFP via Getty

ChatGPT continues to steal the spotlight, more than a year after its public debut.

The artificial intelligence (AI) chatbot was released as a free-to-use tool in November 2022 by tech company OpenAI in San Francisco, California. Two months later, ChatGPT had already been listed as an author on a handful of research papers.

ChatGPT one year on: who is using it, how and why?

Academic publishers scrambled to announce policies on the use of ChatGPT and other large language models (LLMs) in the writing process. By last October, 87 of 100 top scientific journals had provided guidance to authors on generative AI, which can create text, images and other content, researchers reported on 31 January in the The BMJ¹.

But that’s not the only way in which ChatGPT and other LLMs have begun to change scientific writing. In academia’s competitive environment, any tool that allows researchers to “produce more publications is going to be a very attractive proposition”, says digital-innovation researcher Savvas Papagiannidis at Newcastle University in Newcastle upon Tyne, UK.

Generative AI is continuing to improve — so publishers, grant-funding agencies and scientists must consider what constitutes ethical use of LLMs, and what over-reliance on these tools says about a research landscape that encourages hyper-productivity.

Are scientists routinely using LLMs to write papers?

Before its public release, ChatGPT was not nearly as user-friendly as it is today, says computer scientist Debora Weber-Wulff at the HTW Berlin University of Applied Sciences. “The interfaces for the older GPT models were something that only a computer scientist could love.”

In the past, researchers typically needed specialized expertise to use advanced LLMs. Now, “GPT has democratized that to some degree”, says Papagiannidis.

This democratization has catalysed the use of LLMs in research writing. In a 2023 Nature survey of more than 1,600 scientists, almost 30% said that they had used generative AI tools to help write manuscripts, and about 15% said they had used them to help write grant applications.

And LLMs have many other uses. They can help scientists to write code, brainstorm research ideas and conduct literature reviews. LLMs from other developers are improving as well, such as Google’s Gemini and Claude 2 by Anthropic, an AI company in San Francisco. Researchers with the right skills can even develop their own personalized LLMs that are fine-tuned to their writing style and scientific field, says Thomas Lancaster, a computer scientist at Imperial College London.

What are the benefits for researchers?

About 55% of the respondents to the Nature survey felt that a major benefit of generative AI is its ability to edit and translate writing for researchers whose first language is not English. Similarly, in a poll by the European Research Council (ERC), which funds research in Europe, 75% of more than 1,000 ERC-grant recipients felt that generative AI will reduce language barriers in research by 2030, according to a report released in December².

Of the ERC survey respondents, 85% thought that generative AI could take on repetitive or labour-intensive tasks, such as literature reviews. And 38% felt that generative AI will promote productivity in science, such as by helping researchers to write papers at a faster pace.

What are the downsides?

Although ChatGPT’s output can be convincingly human-like, Weber-Wulff warns that LLMs can still make language mistakes that readers might notice. That’s one of the reasons she advocates for researchers to acknowledge LLM use in their papers. Chatbots are also notorious for generating fabricated information, called hallucinations.

How ChatGPT and other AI tools could disrupt scientific publishing

And there is a drawback to the productivity boost that LLMs might bring. Speeding up the paper-writing process could increase throughput at journals, potentially stretching editors and peer reviewers even thinner than they already are. “With this ever-increasing number of papers — because the numbers are going up every year — there just aren’t enough people available to continue to do free peer review for publishers,” Lancaster says. He points out that alongside researchers who openly use LLMs and acknowledge it, some quietly use the tools to churn out low-value research.

It’s already difficult to sift through the sea of published papers to find meaningful research, Papagiannidis says. If ChatGPT and other LLMs increase output, this will prove even more challenging.

“We have to go back and look at what the reward system is in academia,” Weber-Wulff says. The current ‘publish or perish’ model rewards researchers for constantly pushing out papers. But many people argue that this needs to shift towards a system that prioritizes quality over quantity. For example, Weber-Wulff says, the German Research Foundation allows grant applicants to include only ten publications in a proposal. “You want to focus your work on getting really good, high-level papers,” she says.

Where do scientific publishers stand on LLM use?

According to the study in The BMJ, 24 of the 100 largest publishers — collectively responsible for more than 28,000 journals — had by last October provided guidance on generative AI¹. Journals with generative-AI policies tend to allow some use of ChatGPT and other LLMs, as long as they’re properly acknowledged.

Springer Nature, for example, states that LLM use should be documented in the methods or another section of the manuscript, a guideline introduced in January 2023. Generative AI tools do not, however, satisfy criteria for authorship, because that “carries with it accountability for the work, and AI tools cannot take such responsibility”. (Nature’s news team is editorially independent of its publisher, Springer Nature.)

Enforcing these rules is easier said than done, because undisclosed AI-generated text can be difficult for publishers and peer reviewers to spot. Some sleuths have caught it through subtle phrases and mistranslations. Unlike cases of plagiarism, in which there is clear source material, “you can’t prove that anything was written by AI”, Weber-Wulff says. Despite researchers racing to create LLM-detection tools, “we haven’t seen one that we thought produced a compelling enough result” to screen journal submissions, says Holden Thorp, editor in chief of the Science family of journals.

What about other uses?

Although as of November, the American Association for the Advancement of Science — which publishes Science — allows for some disclosed use of generative AI in the preparation of manuscripts, it still bans the use of LLMs during peer review, Thorp says. This is because he and others at Science want reviewers to devote their full attention to the manuscript being assessed, he adds. Similarly, Springer Nature’s policy prohibits peer reviewers from uploading manuscripts into generative-AI tools.

Scientific sleuths spot dishonest ChatGPT use in papers

Some grant-funding agencies, including the US National Institutes of Health and the Australian Research Council, forbid reviewers from using generative AI to help examine grant applications because of concerns about confidentiality (grant proposals are treated as confidential documents, and the data entered into public LLMs could be accessed by other people). But the ERC Scientific Council, which governs the ERC, released a statement in December recognizing that researchers use AI technologies, along with other forms of external help, to prepare grant proposals. It said that, in these cases, authors must still take full responsibility for their work.

“Many organizations come out now with very defensive statements” requiring authors to acknowledge all use of generative AI, says ERC Scientific Council member Tom Henzinger, a computer scientist at the Institute of Science and Technology Austria in Klosterneuburg.

To him, ChatGPT seems no different from running text by a colleague for feedback. “Use every resource at your disposal,” Henzinger says.

Regardless of the ever-changing rules around generative AI, researchers will continue to use it, Lancaster says. “There is no way of policing the use of technology like ChatGPT.”

[ad_2]

Source link

February 28, 2024
Why a publisher retracted abortion-pill studies cited in a case set for the Supreme Court

[ad_1]

A physician at a clinic in New Mexico watches as a person takes the abortion pill mifepristone in 2023.Credit: Evelyn Hockstein/Reuters

Early this month, a scientific publisher retracted two studies¹^,² cited by a federal judge in Texas when he ruled that the abortion pill mifepristone should be taken off the market, suggesting that the drug causes a burden on the public-health system. It also retracted a third³ that surveyed abortion providers in Florida, linking them to malpractice and disciplinary issues. According to Sage Publications, the first two papers had problems with study design and methodology and errors in data analysis. And all three included unsupported assumptions and misleading data presentations. In addition, the studies’ authors, many of whom are affiliated with anti-abortion organizations, failed to declare conflicts of interest, Sage said in its retraction notice.

Nature spoke to the researcher who contacted Sage with concerns about the papers, as well as to reproductive-health specialists to learn about the perceived issues that triggered the papers’ retractions. They praise the retractions, but say that there are many similar publications alleging the harms of abortion that have yet to be addressed.

James Studnicki, the lead author of the three papers and director of data analytics at the Charlotte Lozier Institute (CLI) in Arlington, Virginia, which describes itself as a pro-life research organization, said in a statement that there is “no legitimate reason for Sage’s retractions”, and that the authors “fully complied with Sage’s conflict disclosure requirements” by reporting their affiliations and CLI funding. The authors will be taking legal action against Sage, according to Studnicki.

Papers questioned

Chris Adkins, a pharmaceutical scientist at South University in Savannah, Georgia, first came across one of the Sage papers after it was cited in April 2023 in a ruling by Matthew Kacsmaryk in the US District Court for the Northern District of Texas. Kacsmaryk pointed to the study, published in 2021¹, as evidence that mifepristone-induced abortions lead to an elevated incidence of emergency-room (ER) visits.

“I found enough issues in the paper that I felt compelled to reach out to the journal,” Adkins says — especially given its impact.

Activists protest against a ruling restricting the availability of the abortion drug mifepristone in Texas.Credit: Olga Fedorova/SOPA Images/Shutterstock

The Texas ruling has since been appealed, and the lawsuit has wended its way to the US Supreme Court, which will hear arguments in late March about whether mifepristone use should be restricted nationwide.

After hearing concerns about the 2021 paper, Sage began an investigation. Two more papers by some of the same authors were included in the review, and the publisher enlisted independent experts to examine the science behind the studies.

The 2021 paper compares the number of ER visits in the 30 days after a surgical abortion with those after a medication-induced abortion, using data from Medicaid, a US government programme that provides health insurance to people with limited resources. The conclusion, now retracted, was that medication-induced abortions were linked to more visits.

One problem, Adkins says, is that the study claims that the incidence of visits after any type of induced abortion is increasing year on year, without comparing the trend with that in overall ER visits. If overall ER visits were increasing owing to, say, a rise in Medicaid use, the trend could not be attributed to abortions becoming riskier.

Abortion-pill ruling threatens FDA’s authority, say drug firms

The authors pointed Nature to a rebuttal letter they publicly released after Sage’s investigation, in response to a request for comment. They deny that the study’s focus was on comparing people who had an abortion with those who didn’t. One conclusion listed in the paper begins: “The incidence and per-abortion rate of ER visits following any induced abortion are growing”.

Another issue raised by researchers is that the study uses ER visits as a proxy for abortion-related complications, says Ushma Upadhyay, a reproductive-health specialist at the University of California, San Francisco. “We know that many people go to an emergency department because they live too far from the abortion provider,” she says, and they want someone to check any bleeding they might experience after taking mifepristone. Many studies⁴ have shown that mifepristone is safe, and that bleeding is a normal, short-lived side effect of taking it — not a complication.

In their rebuttal letter, the authors quote from their 2021 paper, saying that ER visits are “particularly insightful” events to use when comparing the relative safety of chemical and surgical abortions. “Adverse events following a mifepristone abortion are more likely to be experienced at home in the absence of a physician, increasing the likelihood of an ER visit,” they add.

Although Sage did not publicly release the findings of its independent reviewers, the authors’ rebuttal letter gives insight into other problems that the experts flagged.

US lawsuit threatens access to abortion drug: the science behind the case

One of the papers, published in 2019³, investigates the characteristics of physicians who provide abortions in the state of Florida. It says that nearly half of the abortion providers that the researchers evaluated had at least one malpractice claim, public complaint, disciplinary action or criminal charge against them, without providing any comparison with the overall rate of such claims in the general physician population. According to the rebuttal letter, two independent reviewers noted that, because abortion providers do not have to advertise their services publicly or necessarily register with the state, the cohort investigated by the authors might be biased in some unknown direction.

The authors say in their letter that the paper made no claims that the sample was statistically representative or could be generalized to other states.

When asked by Nature how the papers made it through review, a Sage spokesperson responded that the publisher relies on journal editors to make individual decisions on submitted works based on the evaluations of peer reviewers. In its retraction notice, Sage said that it discovered one peer reviewer who had evaluated the three papers was affiliated with an anti-abortion organization.

Roadblocks to retractions

Upadhyay was surprised — and relieved — to hear the news of the retractions. It’s difficult for publishers to retract these types of articles, she says. “In the past, we’ve seen that anti-abortion researchers have threatened lawsuits against the publishers.”

Chelsea Polis, an epidemiologist at the research organization Population Council in New York City, points to a meta-analysis published in the British Journal of Psychiatry⁵ as an example. Many scholars, including Polis and her colleagues, have published letters pointing out concerns about the methods used in the paper, which concluded that there’s an increased risk of mental-health problems after an abortion.

‘We are bearing witness’: Health researchers navigate a post-Roe world

An investigation by The BMJ last year reported that even after an internal panel appointed by the journal recommended that the article should be retracted, the journal declined to do so. Members of that panel resigned from the journal’s board as a result and suggested that the publisher, the Royal College of Psychiatrists in London, fears being sued. The author, Priscilla Coleman, a psychologist retired from Bowling Green State University in Ohio, threatened legal action after she was notified that the paper was being investigated.

Coleman did not respond to Nature’s request for comment.

Contacted by Nature, the Royal College of Psychiatrists did not comment on what motivated its decision. Instead, it pointed to a 2023 statement indicating that “the widely available public debate on the paper, including the letters of complaint already available alongside the article online”, made it unnecessary to retract the study. According to a commentary published today in The BMJ⁶, the paper has been cited in 25 court cases, including the ruling by Kacsmaryk, as well as in 14 parliamentary hearings in 6 countries.

Polis, who has herself been sued because of another complaint she lodged that led to a paper being retracted, says that these legal threats discourage academics from speaking out against problematic papers. “At least in my field of sexual and reproductive health, I don’t think enough feel compelled to action,” she adds. “At present, there is a lot of risk in taking on this kind of work, and very few advantages.”

[ad_2]

Source link

February 27, 2024
How a Small Iowa Newspaper’s Website Became an AI-Generated Clickbait Factory

[ad_1]

In his spare time, Tony Eastin likes to dabble in the stock market. One day last year, he Googled a pharmaceutical company that seemed like a promising investment. One of the first search results Google served up on its news tab was listed as coming from the Clayton County Register, a newspaper in northeastern Iowa. He clicked, and read. The story was garbled and devoid of useful information—and so were all the other finance-themed posts filling the site, which had absolutely nothing to do with northeastern Iowa. “I knew right away there was something off,” he says. There’s plenty of junk on the internet, but this struck Eastin as strange: Why would a small midwestern paper churn out crappy blog posts about retail investing?

Eastin was primed to find online mysteries irresistible. After years in the US Air Force working on psychological warfare campaigns he had joined Meta where he investigated nastiness ranging from child abuse to political influence operations. Now he was between jobs, and welcomed a new mission. So Eastin reached out to Sandeep Abraham, a friend and former Meta colleague who previously worked in Army intelligence and for the NSA, and suggested they start digging.

What the pair uncovered provides a snapshot of how generative AI is enabling deceptive new online business models. Networks of websites crammed with AI-generated clickbait are being built by preying on the reputations of established media outlets and brands. These outlets prosper by confusing and misleading audiences and advertisers alike, “domain squatting” on URLs that once belonged to more reputable organizations. The scuzzy site Eastin was referred to no longer belonged to the newspaper whose name it still traded in the name of.

Although Eastin and Abraham suspect that the network the Register’s old site is now part of was created with straightforward money-making goals, they fear that more malicious actors could use the same sort of tactics to push misinformation and propaganda into search results. “This is massively threatening,” Abraham says. “We want to raise some alarm bells.” To that end, the pair have released a report on their findings and plan to release more as they dig deeper into the world of AI clickbait, hoping their spare-time efforts can help draw awareness to the issue from the public or lawmakers.

Faked News

The Clayton County Register was founded in 1926 and covered the small town of Ekader, Iowa, and wider Clayton County, which nestle against the Mississippi River in the state’s northeast corner. “It was a popular paper,” says former coeditor Bryce Durbin, who describes himself as “disgusted” by what’s now published at its former web address, claytoncountyregister.com. (The real Clayton County Register merged in 2020 with The North Iowa Times to become the Times-Register which publishes at a different website. It’s not clear how the paper lost control of its web domain; the Times-Register did not return requests for comment.)

As Eastin discovered when trying to research his pharma stock, the site still brands itself as the Clayton County Register but no longer offers local news and is instead a financial news content mill. It publishes what appear to be AI-generated articles about the stock prices of public utility companies and Web3 startups, illustrated by images that are also apparently AI-generated.

“Not only are the articles we looked at generated by AI, but the images included in each article were all created using diffusion models,” says Ben Colman, CEO of deepfake detection startup Reality Defender, which ran an analysis on several articles at WIRED’s request. In addition to that confirmation, Abraham and Eastin noticed that some of the articles included text admitting their artificial origins. “It’s important to note that this information was auto-generated by Automated Insights,” some of the articles stated, name-dropping a company that offers language-generation technology.

[ad_2]

Source link

February 26, 2024
Why it would be a dangerous folly to end US–China science pact

[ad_1]

China and the United States cooperated to ensure that miniature neutron source reactors can run on low-enriched uranium.Credit: Xinhua/Cai Yang/Alamy

Two things can be said of the continuing delay to renewing the US–China Science and Technology Cooperation Agreement. The good news is that the two sides are still talking about continuing with the landmark 45-year-old agreement, which has yielded historic levels of research collaboration and student exchanges between the two countries. The bad news is that one or both sides could still walk away. This would be catastrophic. Wisdom and forward thinking must prevail.

Before China and the United States established diplomatic relations on 1 January 1979, there was little or no formal relationship between the two nations, and high levels of mistrust. Science cooperation was identified as offering a relatively swift way to break the ice and begin establishing people-to-people contacts. Then-US president Jimmy Carter and China’s premier at the time, Deng Xiaoping, signed the science agreement before the month was out, on 31 January.

US and China likely to delay renewal of key science pact again

Admittedly, the two countries’ motivations for pursuing scientific cooperation were different. For China, the decision was development-led. The nation was far from the research-driven power that it is now. Today, it boasts some 3,000 higher-education institutions; back then, its annual per-capita income stood at less than US$200. China’s leaders wanted to learn how to build a world-class higher education system, as well as how they might use research to boost economic growth, and, by extension, living standards, as Julian Gewirtz, a historian of China–US economic-research ties, writes in Unlikely Partners (2017). The United States also had a political goal: to steer China away from the orbit of the Soviet Union during the ongoing cold war.

Today, although the two countries can hardly be described as ‘best friends forever’, the fruits of their collaboration are clear. Some 3 million Chinese students have studied at universities in the United States since the agreement was brokered. In 2021, US universities awarded more than 8,000 doctorates to students from China, out of a total of around 25,000 international doctorates. Each country is the other’s biggest research partner, by a considerable margin.

Relations took a negative turn during Donald Trump’s US presidency, from 2017 to 2021. After the start of the COVID-19 pandemic, rhetoric harshened significantly, and this was followed by an erosion of trade and diplomatic links. In the realms of research and higher education, a narrative took hold that there is something inherently suspicious about cooperation between US and Chinese researchers — with an emphasis on known threats such as spying and intellectual-property theft. This has clearly affected collaborations, but has also had a broader reach. There has been surveillance of some innocent researchers. And Florida’s decision to stop universities hiring researchers from China (as well as from Iran and a handful of other countries) would not have seemed out of place during the cold war.

The United States has not been alone in initiating decoupling efforts. From March 2020, China’s government adopted a policy whereby its researchers would no longer be encouraged to publish in international journals. China’s leadership has also taken to talking more and more about self-reliance, one implication of which is less and less need for collaborative effort.

Mutual benefits

John Holdren, a physicist at Harvard University in Cambridge, Massachusetts, was science adviser to former US president Barack Obama and, in 2011, he and Wan Gang, China’s then minister of science and technology, renewed the US–China science pact. That agreement was designed to ensure that the benefits would be mutual, Holdren tells Nature. Those benefits are both national and global.

Collaboration between the two countries on environmental protection includes projects to monitor and improve air and water quality, as well as watershed protection, and projects to reduce electronic waste — benefiting both countries in different ways. The US Environmental Protection Agency has called its relationship with China “one of its most significant”.

In 2011, China’s then science minister Wan Gang and then US science adviser John Holdren renewed the science pact originally agreed between former Chinese premier Deng Xiaoping and former US president Jimmy Carter (depicted in black and white photograph).Credit: Nature and Science/Alamy

When it comes to global challenges, researchers in China, the United States and Europe are cooperating extensively on studying the role of nature in human prosperity¹^,². This evolving body of work is foundational to ongoing efforts to incorporate nature into how economies are valued.

Another notable but little-known project aims to reduce the risk of nuclear proliferation. Since 2009, China and the United States have been working together to convert a type of nuclear research reactor called a miniature neutron source reactor so that instead of using highly enriched, weapons-grade uranium as fuel, it runs on low-enriched uranium — which cannot be used in nuclear weapons. China has supplied this type of reactor to a number of countries, including Iran, Nigeria and Pakistan. In a small way, this cooperation has contributed to a safer world.

‘This is exclusion’: Florida law restricts hiring of researchers from seven countries

And then there’s climate change. After a period of silence that began in 2022, the two countries began talking again last year, thanks in no small measure to the long-standing relationship between their then climate envoys, John Kerry and Xie Zhenhua. Last year, California made an agreement with China pledging to cut carbon emissions and transition away from using fossil fuels. Both Kerry and Zhenhua are moving on to new roles, and the legacy of their diplomatic efforts risks being undermined if scientists in the two nations cannot maintain their research ties.

Germany’s handling of its research relations with China could offer lessons. Last month, the German Academic Exchange Service published some sensible recommendations that balance the risks of such collaborations with the benefits.

The document acknowledges the benefits that have come from closer ties, while advocating what it calls a “realpolitik approach” to future links — one based on practical objectives, rather than ideology. Ultimately, it says that universities should be the ones to decide what is mutually beneficial in this regard, while taking the necessary precautions to protect against possible harm.

Risk management

There are, of course, always risks when researchers from different political systems collaborate. And it shouldn’t surprise anyone that big powers spy on each other, says Holdren. But, as with most applications of science in public affairs, from nanotechnology to nuclear energy, the answer to handling risks is to assess them, manage them and mitigate them — always using rigorously tested scientific knowledge.

After 45 years of scientific cooperation, the United States and China risk veering off course. It would be a dangerous folly to bring an end to research cooperation that has such potential to help meet the many challenges faced by China, the United States and the world. In 1979, scientists broke the ice at a time of great tension. As tensions rise once again, researchers could be the foot in the door that keeps communications open.

[ad_2]

Source link

February 26, 2024
‘All of Us’ genetics chart stirs unease over controversial depiction of race

[ad_1]

Geneticists have become concerned about how the research they publish could be interpreted incorrectly by the public.Credit: Tek Image/Science Photo Library

Some geneticists have expressed their unease about a figure in a high-profile Nature paper that was published earlier this week¹, noting that it could be misinterpreted as reinforcing racist beliefs. The figure has reignited a long-standing debate among geneticists about how best to discuss and depict race, ethnicity and genomic ancestry, given how these terms can be misinterpreted and weaponized by extremists.

Ambitious survey of human diversity yields millions of undiscovered genetic variants

“The problem is, a lot of people will see figures like this as supporting a viewpoint” that race and ethnicity are closely aligned with genetics, says Ewan Birney, deputy director-general of the European Molecular Biology Laboratory in Cambridgeshire, UK. “And then they build castles in the air from all this.”

Alexander Bick, a physician and geneticist at Vanderbilt University Medical Center in Nashville, Tennessee, who co-authored the paper in which the figure appears, acknowledged in an e-mail to Nature’s news team that “it’s clear that the figure fell short of our intended goal for this paper”. (Nature’s news team is editorially independent of its journal team.) But Bick defended the analysis, noting that it is a “faithful representation of the patterns that exist in the data that is consistent with representations in other similar studies” and that he is not planning to submit a correction to remove the plot.

Stirring debate

The paper is part of a larger package of articles published on 19 February that detail the progress and initial analyses of the All of Us programme, run by the US National Institutes of Health in Bethesda, Maryland. The programme aims to tackle inequities in genetics research by assembling detailed genomic and health profiles for a diverse group of one million people in the United States by the end of 2026.

Within hours of the papers’ publication, population geneticist Jonathan Pritchard at Stanford University in California posted on the social-media platform X (formerly Twitter) to share his concerns about the figure, which is intended to showcase the diversity of the first 250,000 genomes included in the All of Us database. The chart uses an algorithm called UMAP to visualize genetic relationships and participants’ self-described race and ethnicity. But a problem with using UMAP, Pritchard wrote, is that it can exaggerate the distinctiveness of populations and fail to represent their intermixing properly.

In reality, “genetic variation is a continuum, and thus genetic ancestry cannot be objectively carved out into discrete groups”, says Roshni Patel, a statistical geneticist who works with Pritchard at Stanford University.

An excerpt from a figure in a Nature paper that some geneticists say could be misinterpreted to reinforce racist beliefs. Source: Ref 1.

To a layperson, the chart shows several distinct colourful blobs that could be misinterpreted as supporting genetic essentialism — the pseudoscientific belief that racial or ethnic groups are distinct genetic categories, and that individuals of the same group are genetically similar, Birney says.

That is the opposite of what the data show, Bick says. “Our analysis reaffirms that race and ethnicity are social constructs that do not have a basis in genetics”.

A few researchers have publicly called on Nature’s journal team to retract the paper because of the potentially misleading nature of the figure. Most geneticists contacted by Nature’s news team did not think retraction was necessary, but said they hoped that the authors would acknowledge the concerns. A spokesperson for the journal says that Nature’s editors are “aware of the discussions that are taking place and are in contact with the authors”.

Public interpretation

Geneticists have become particularly sensitive to how analyses they publish are used or interpreted by the public. In 2022, an 18-year-old gunman in Buffalo, New York, attempted to justify killing 10 Black people at a grocery shop with a 180-page manifesto that included several citations and figures from genetics papers.

Facing up to injustice in genome science

This ignited a debate among geneticists about their responsibility to ensure that their research does not spread in pseudoscientific circles and is not used to justify violence. “The language of some of this work showing up in other people’s violent manifestos is a very sobering reality that geneticists are having to reconcile with,” says Nicole Iturriaga, a political sociologist at the University of California, Irvine, who has studied how genetics research is co-opted by far-right groups.

But Birney says he isn’t convinced there is a “neat and pretty way of representing relationships between humans”. Human genetic relationships are best described by family trees, and any effort to squash the hundreds of thousands of people included in a data set into a single chart will leave out important context, he says. It doesn’t help that every country uses the terms ‘race’ and ‘ethnicity’ in its own way, which makes the concepts harder for the public to understand, he adds.

That doesn’t mean that geneticists should stop using charts in their manuscripts; it means that extra care is necessary to ensure that data are communicated responsibly, says Anna Lewis, a specialist in the ethical implications of genomics at Brigham and Women’s Hospital in Boston, Massachusetts. Patel adds that the All of Us researchers could have more closely followed the recommendations of a 2023 report issued by the US National Academies of Sciences, Engineering, and Medicine (NASEM) about best practices for using population descriptors in genetics research. Bick responds that the manuscript was written in 2022, before the NASEM report was published.

Despite the charged and public nature of the criticism, Iturriaga says that this is a healthy debate for the field to have. And it’s important that geneticists think about the potential for misinterpretation before it’s too late, she adds.

[ad_2]

Source link

February 23, 2024
China conducts first nationwide review of retractions and research misconduct

[ad_1]

The reputation of Chinese science has been “adversely affected” by the number of retractions in recent years, according to a government notice.Credit: Qilai Shen/Bloomberg/Getty

Chinese universities are days away from the deadline to complete a nationwide audit of retracted research papers and probe of research misconduct. By 15 February, universities must submit to the government a comprehensive list of all academic articles retracted from English- and Chinese-language journals in the past three years. They need to clarify why the papers were retracted and investigate cases involving misconduct, according to a 20 November notice from the Ministry of Education’s Department of Science, Technology and Informatization.

The government launched the nationwide self-review in response to Hindawi, a London-based subsidiary of the publisher Wiley, retracting a large number of papers by Chinese authors. These retractions, along with those from other publishers, “have adversely affected our country’s academic reputation and academic environment”, the notice states.

A Nature analysis shows that last year, Hindawi issued more than 9,600 retractions, of which the vast majority — about 8,200 — had a co-author in China. Nearly 14,000 retraction notices, of which some three-quarters involved a Chinese co-author, were issued by all publishers in 2023.

This is “the first time we’ve seen such a national operation on retraction investigations”, says Xiaotian Chen, a library and information scientist at Bradley University in Peoria, Illinois, who has studied retractions and research misconduct in China. Previous investigations have largely been carried out on a case-by-case basis — but this time, all institutions have to conduct their investigations simultaneously, says Chen.

Tight deadline

The ministry’s notice set off a chain of alerts, cascading to individual university departments. Bulletins posted on university websites required researchers to submit their retractions by a range of dates, mostly in January — leaving time for universities to collate and present the data.

Although the alerts included lists of retractions that the ministry or the universities were aware of, they also called for unlisted retractions to be added.

More than 10,000 research papers were retracted in 2023 — a new record

According to Nature’s analysis, which includes only English-language journals, more than 17,000 retraction notices for papers published by Chinese co-authors have been issued since 1 January 2021, which is the start of the period of review specified in the notice. The analysis, an update of one conducted in December, used the Retraction Watch database, augmented with retraction notices collated from the Dimensions database, and involved assistance from Guillaume Cabanac, a computer scientist at the University of Toulouse in France. It is unclear whether the official lists contain the same number of retracted papers.

Regardless, the timing to submit the information will be tight, says Shu Fei, a bibliometrics scientist at Hangzhou Dianzi University in China. The ministry gave universities less than three months to complete their self-review — and this was cut shorter by the academic winter break, which typically starts in mid-January and concludes after the Chinese New Year, which fell this year on 10 February.

“The timing is not good,” he says. Shu expects that universities are most likely to submit only a preliminary report of their researchers’ retracted papers included on the official lists.

But Wang Fei, who studies research-integrity policy at Dalian University of Technology in China, says that because the ministry has set a deadline, universities will work hard to submit their findings on time.

Researchers with retracted papers will have to explain whether the retraction was owing to misconduct, such as image manipulation, or an honest mistake, such as authors identifying errors in their own work, says Chen: “In other words, they may have to defend themselves.” Universities then must investigate and penalize misconduct. If a researcher fails to declare their retracted paper and it is later uncovered, they will be punished, according to the ministry notice. The cost of not reporting is high, says Chen. “This is a very serious measure.”

It is not known what form punishment might take, but in 2021, China’s National Health Commission posted the results of its investigations into a batch of retracted papers. Punishments included salary cuts, withdrawal of bonuses, demotions and timed suspensions from applying for research grants and rewards.

The notice explicitly states that the first corresponding author of a paper is responsible for submitting the response. This requirement will largely address the problem of researchers shirking responsibility for collaborative work, says Li Tang, a science- and innovation-policy researcher at Fudan University in Shanghai, China. The notice also emphasizes due process, says Tang. Researchers alleged to have committed misconduct have a right to appeal during the investigation.

The notice is a good approach for addressing misconduct, says Wang. Previous efforts by the Chinese government have stopped at issuing new research-integrity guidelines that were poorly implemented, she says. And when government bodies did launch self-investigations of published literature, they were narrower in scope and lacked clear objectives. This time, the target is clear — retractions — and the scope is broad, involving the entire university research community, she says.

“Cultivating research integrity takes time, but China is on the right track,” says Tang.

What next

It is not clear what the ministry will do with the flurry of submissions. Wang says that, because the retraction notices are already freely available, publicizing the collated lists and underlying reasons for retraction could be useful. She hopes that a similar review will be conducted every year “to put more pressure” on authors and universities to monitor research integrity.

What happens next will reveal how seriously the ministry regards research misconduct, says Shu. He suggests that, if the ministry does not take further action after Chinese New Year, the notice could be an attempt to respond to the reputational damage caused by the mass retractions last year.

The ministry did not respond to Nature’s questions about the misconduct investigation.

Chen says regardless of what the ministry does with the information, the reporting process itself will help to curb misconduct because it is “embarrassing to the people in the report”.

But it might primarily affect researchers publishing in English-language journals. Retraction notices in Chinese-language journals are rare.

[ad_2]

Source link

February 12, 2024
How journals are fighting back against a wave of questionable images

[ad_1]

Journals are making an effort to detect manipulated images of the gels used to analyse proteins and DNA.Credit: Shutterstock

It seems that every month brings a fresh slew of high-profile allegations against researchers whose papers — some of them years old — contain signs of possible image manipulation.

Scientist sleuths are using their own trained eyes, along with commercial software based on artificial intelligence (AI), to spot image duplication and other issues that might hint at sloppy record-keeping or worse. They are bringing these concerns to light in places like PubPeer, an online forum featuring many new posts every day flagging image concerns.

Some of these efforts have led to action. Last month, for example, the Dana-Farber Cancer Institute (DFCI) in Boston, Massachusetts, said that it would ask journals to retract or correct a slew of papers authored by its staff members. The disclosure came after an observer raised concerns about images in the papers. The institute says it is continuing to investigate the concerns.

That incident was just one of many. In the face of public scrutiny, academic journals are increasingly adopting tricks and tools, including commercial AI-based systems, to spot problematic imagery before, rather than after, publication. Here, Nature reviews the problem and how publishers are attempting to tackle it.

What sorts of imagery problem are being spotted?

Questionable image practices include the use of the same data across several graphs, the replication of photos or portions of photos, and the deletion or splicing of images. Such issues can indicate an intent to mislead, but can also result from an innocent attempt to improve a figure’s aesthetics, for example. Nonetheless, even innocent mistakes can be damaging to the integrity of science, experts say.

How prevalent are these issues, and are they on the rise?

The precise number of such incidents is unknown. A database maintained by the website Retraction Watch lists more than 51,000 documented retractions, corrections or expressions of concern. Of those, about 4% flag a concern about images.

Meet this super-spotter of duplicated images in science papers

One of the largest efforts to quantify the problem was carried out by Elisabeth Bik, a scientific image sleuth and consultant in San Francisco, California, and her colleagues¹. They examined images in more than 20,000 papers that were published between 1995 and 2014. Overall, they found that nearly 4% of the papers contained problematic figures. The study also revealed an increase in inappropriate image duplications starting around 2003, probably because digital photography made it easier to alter photos, Bik says.

Modern papers also contain more images than do those from decades ago, notes Bik. “Combine all of this with many more papers being published per day compared to ten years ago, and the increased pressure put on scientists to publish, and there will just be many more problems that can be found.”

The high rate of reports of image issues might also be driven by “a rise in whistleblowing because of the global community’s increased awareness of integrity issues”, says Renee Hoch, who works for the PLOS Publication Ethics team in San Francisco, California.

What happened at the Dana-Farber Cancer Institute?

In January, biologist and investigator Sholto David, based in Pontypridd, UK, blogged about possible image manipulation in more than 50 biology papers published by scientists at the DFCI, which is affiliated with Harvard University in Cambridge, Massachusetts. Among the authors were DFCI president Laurie Glimcher and her deputy, William Hahn; a DFCI spokesperson said they are not speaking to reporters. David’s blog highlighted what seemed to be duplications or other image anomalies in papers spanning almost 20 years. The post was first reported by The Harvard Crimson.

The DFCI, which had already been investigating some of these issues, is seeking retractions for several papers and corrections for many others. Barrett Rollins, the DFCI’s research-integrity officer, says that “moving as quickly as possible to correct the scientific record is important and a common practice of institutions with strong research integrity”.

“It bears repeating that the presence of image duplications or discrepancies in a paper is not evidence of an author’s intent to deceive,” she adds.

What are journals doing to improve image integrity?

In an effort to reduce publication of mishandled images, some journals, including the Journal of Cell Science, PLOS Biology and PLOS ONE, either require or ask that authors submit raw images in addition to the cropped or processed images in their figures.

Many publishers are also incorporating AI-based tools including ImageTwin, ImaCheck and Proofig into consistent or spot pre-publication checks. The Science family of journals announced in January it is now using Proofig to screen all its submissions. Holden Thorp, editor in chief of the Science family of journals, says Proofig has spotted things that led editors to decide against publishing papers. He says authors are usually grateful to have their errors identified.

What kinds of issues do these AI-based systems flag?

All these systems can, for example, quickly detect duplicates of images within the same paper, even if those images have been rotated, stretched or cropped or had their colour altered.

Different systems have different merits. Proofig, for example, can spot splices created by chopping out or stitching together portions of images. ImageTwin, says Bik, has the advantage of allowing users to cross-check an image against a large dataset of other papers. Some publishers, including Springer Nature, are developing their own AI image integrity software. (Nature’s news team is editorially independent of its publisher, Springer Nature.)

Many of the errors flagged by AI tools seem to be innocent. In a study of more than 1,300 papers submitted to 9 American Association for Cancer Research journals in 2021 and early 2022, Proofig flagged 15% as having possible image duplications that required follow-up with authors. Author responses indicated that 28% of the 207 duplications were intentional — driven, for example, by authors using the same image to illustrate multiple points. Sixty-three per cent were unintentional mistakes.

How well do these AI systems work?

Users report that AI-based systems definitely make it faster and easier to spot some kinds of image problems. The Journal of Clinical Investigation trialled Proofig from 2021–2022 and found that it tripled the proportion of manuscripts with potentially problematic images, from 1% to 3%².

But they are less adept at spotting more complex manipulations, says Bik, or AI-generated fakery. The tools are “useful to detect mistakes and low-level integrity breaches, but that is but one small aspect of the bigger issue,” agrees Bernd Pulverer, chief editor of EMBO Reports. “The existing tools are at best showing the tip of an iceberg that may grow dramatically, and current approaches will soon be largely obsolete.”

Are pre-publication checks stemming image issues?

A combination of expert teams, technology tools and increased vigilance seems to be working — for the time being. “We have applied systematic screening now for over a decade and for the first time see detection rates decline,” says Pulverer.

But as image manipulation gets more sophisticated, catching it will become ever harder, he says. “In a couple of years all of our current image integrity screening will still be useful for filtering out mistakes, but certainly not for detecting fraud,” Pulverer says.

How can image manipulation best be tackled in the long run?

Ultimately, stamping out image manipulation will involve complex changes to how science is done, says Bik, with more focus on rigor and reproducibility, and repercussions for bad behaviour. “There are too many stories of bullying and highly demanding PIs spending too little time in their labs, and that just creates a culture where cheating is ok,” she says. “This needs to change.”

[ad_2]

Source link

February 12, 2024