A UB professor dismantles the scientific study that there may be identical fingerprints

2024-01-12 06:01:11

With very rare exceptions, our fingerprints are unique. No two are identical – not even among those of the same person – and therefore they are incomparable to each other, as considered by forensic science. However, a new study questions this scientific principle: a team of iengineers from the American universities of Columbia and Buffalo have created a new fingerprint analysis using artificial intelligence (AI) and have studied 60,000 samples from a public database in the United States. The report, which was supposed to revolutionize forensic science, is also called into question due to the method used and its practical application in legal medicine.

The university research indicated that there are similarities in the fingerprints of different fingers of the same person, although this conclusion cannot be applied, for the moment, to any judicial or police procedure, since it must pass a series of legal and scientific requirements to be accepted by the international community.

One of the most important principles that the legal system has in terms of evidence is that of “plausibility”, as spoints out the professor at the University of Barcelona Manuel Gené Badia, expert in Legal and Forensic Medicine, Forensic Genetics and DNA testing, and Criminal Biological Traces. In this way, a sample, such as a fingerprint found at a crime scene, compared with the same one from a suspect’s finger, will indicate that it statistically may belong to that person, sdepending on the clarity with which the test was taken, but never exactly.

“It is up to the judge, in the case of criminal proceedings, to determine whether that evidence is sufficient to attribute it to a suspect,” the professor tells EL PERIÓDICO. In this way, no matter how much the AI ​​emphasizes that the two fingerprints are linked to the same person, it isIt will be the responsibility of the court to determine the relationship. Furthermore, he emphasizes that there must be a “principle of contradiction” so as not to generate “defenselessness” in the suspect, which is why any expert opinion of evidence or proof, Like a fingerprint, it has to be verified after being analyzed by various experts, who issue their conclusions to the court.

More than one parameter

The professor, who has a degree in Medicine from the University of Barcelona (UB), a doctor in Medicine and Surgery, a diploma in Geriatrics and a specialist in Occupational Medicine and Legal and Forensic Medicine, considers that one should be “prudent” with this type of studies, since many of them are later proven to be inaccurates. For him, what it indicates is that there may be a “link” between random fingerprints that may correspond to fingers of the same subject, although forensic science also observes other parameters in the analysis of this evidence, cHow the position of the sweat glands or the degree of pressure on a surface.

Thus, the traces can be found partial or total and their conservation is determined by the absorbency of the surface on which it is located, of light, humidity or of the reagent for it to appear. If there are two different prints in two scenarios they can be compared, although the fact that they belong to the same person depends on “conditional probability”; that is, that there are other different tests, such as images, other DNA samples or witnesses, that lead us to the same suspect.

The study by the Columbia University and Buffalo University research team, led by Columbia Engineering senior Gabe Guo, fed about 60,000 pairwise fingerprints into an artificial intelligence-based system known as a deep contrastive network. Sometimes the pairs belonged to the same person (but with different fingers) and sometimes they belonged to other subjects.

Another forensic marker

After several months and tests, the AI ​​system detected when seemingly unique fingerprints belonged to the same person and when they did not. According to the study, the precision for a single pair reached 77% but in several pairs the agreement increased. Therefore, researchers believe that this link multiplies forensic efficiency by 10.

The researchers believe that the system’s precision can help prioritize clues in ambiguous situations, although they acknowledge that it is not decisive. Remember that AI uses “a new type of forensic marker” such as “the angles and curvatures of the swirls and loops in the center of the fingerprint” when science focuses more on the so-called ‘minutiae’, which are “the branches and end points in the ridges of fingerprints, patterns used in traditional fingerprint comparison,” according to Guo.

Doubts to publish

The research project was done in collaboration between Professor Hod Lipson’s Creative Machines lab at Columbia Engineering and Professor Wenyao Xu’s Integrated Sensors and Computing lab at the University at Buffalo, and was published this week in the journal Science Advances. ‘.

Related news

In this sense, this publication is not the first choice of researchers, that were rejected by editors of other specific magazines of the forensic scientific community with the argument that it is impossible to detect similarities in two random fingerprints, even from the same person. In addition, the researchers believe that they can find “biases in the analyzed data” which is why they claim to be able to enter millions of results through AI.

This lack of sampling is another of the doubts that the study generates. According to Professor Gené, who is also a member of the Royal Academy of Medicine of Catalonia, there should be matched with more than 60,000 fingerprints, taking into account the volume of the world population, so that it was “statistically significant.” Thus, it could have been established, more solidly, conclusions as specific as determining the existence of similarities between fingerprints of the same person. Sometimes, there are researchers who want to run faster than science itself.