The incredible eyes of Elisabeth Bik: How pattern-matching exposes scientific fraud

As I am in the early- to middle-part of my COVID infection, I've decided that my contribution this week is essentially a redirection to a New York Times Opinion piece by Dr. Elisabeth Bik. Dr. Bik is a microbiologist at Stanford University and the Dutch National Institute for Health with a better-than-average ability to detect patterns. While the NYT article makes it seem that she is the sort who reads scientific papers for fun, her special talent has not made her popular with some of her peers. Her particular skill is identifying image manipulation, whereby photos of blots, agar plates, bacteria from one experiment are flipped, stretched, or cropped to give the appearance of a proven hypothesis or novel finding. Admittedly, Bik doesn't just rely on her eyes for this task. Like other sleuths she utilises software to do some of the work for her, specifically the freely available 29a.ch, but argues that human eyes are still needed to weed out the false positives.

Dr. Bik and compatriots like the pseudonymous Smut Clyde share their findings on twitter and give an eye-opening look into academic misconduct. Below is an example of what Dr. Bik contends with:

Source

Did you notice anything unusual?

Source

It may just look like a series of busy images to most but these were taken from a study that hypothesised that Titanium Oxide nanoparticles had potential as a restorative material in the treatment of cavities as it had no cytotoxic effect.

So, you know, nothing important really. It's just material that could be used for your fillings.

In 2016, Bik and her co-authors Arturo Casadevall and Ferric Fang visually screened 20,621 papers from 40 scientific journals from 1995-2014 and found that 3.8% had problematic figures and half of those were likely to be deliberately falsified. Furthermore, authors who did publish such images were more likely to continue the practice the more they published. Bik and her team found other interesting trends as well: the number of papers with problematic images skyrocketed from 2003. While there was little difference in image duplication between open-access and non-open-access journals, there was a negative correlation between problematic images and impact factor. While PLoS One had the highest number of papers screened, only 4.3% were found to have problematic images compared to 12.4% of papers published in the International Journal of Oncology. Of the problematic papers from PLoS One, China (26.2%) and the USA (40.9%) were often the countries of origin for these papers.

An average biomedical paper can cost between $300,000 and $500,000, so the stakes are high to produce something of note but fabrication can lead to thousands of hours and millions of dollars being wasted in the pursuit of cures or treatments that never worked in the first place. One of the most notable cases of this in recent memory involves neuroscientist Sylvain Lesné's 2006 work in Nature. The Amyloid hypothesis of Alzheimer's Disease posits that the accumulation of amyloid-beta (Aβ) plaques is causal for synaptic loss and neurodegeneration. There are controversies about this theory within Alzheimer's research but it has nonetheless driven drug development for the past twenty years. Lesné and his team claimed to have found a subtype of Aβ which caused dementia in rats. This subtype, subtype 56, was not known to be a main player in the Amyloid Hypothesis, but the paper was subsequently cited in 2300 scholarly articles and led to an injection of almost $300 million in funding from the National Institutes of Health alone into related research. What motivated the whistle-blowing by neuroscientist and physician Matthrew Schrag makes for a fascinating read on its own as it involves questionable FDA practices, short-selling, and how being in the right lab with the right supervisor can push a skilled student into the stratosphere. Bik was consulted on Schrag's findings and found even more evidence of image manipulation.

However, Schrag's efforts placed him under scrutiny and subsequently, work he co-published with his mentor, neuropharmacologist Othman Ghribi, were flagged for containing dubious images. Bik also examined these papers and confirmed the presence of problematic images. Bik herself has had a paper marked with expression of concern.

This is the dark side of science and one that is not entirely rectified by evidence that the original findings are not reproducible. As seen with Alzheimer's, the impact of one falsified study has the power to redirect millions of dollars into similar projects and determine (and derail) the careers of so many young doctors and scientists. A more troubling element is the slow response time on the part of journals to retract or respond to calls for correction when such images are identified; taking into account the millions of students worldwide who may cite such a publication risks entrenching erroneous science and unattainable expectations about experiment outcomes.