Can an algorithm designed by Artificial Intelligence help predict the risk of suffering from psychoses such as schizophrenia or bipolar disorder through fingerprints? Researcher Raymond Salvador confirms this: fingerprints can determine the risk of schizophrenia with a reliability of 70%.
Although it may sound like something in the future, the truth is that the Hospital Sisters Research Foundation (FIDMAG) and the Network Biomedical Research Center in the area of ??Mental Health (CIBERSAM), which depends on the Ministry of Science and Innovation, have been carrying out several years working on a study that demonstrates the usefulness of fingerprints and artificial intelligence in the diagnosis of schizophrenia.
This research delves into the hypothesis that alterations in fingerprints during pregnancy may be related to future mental illnesses, as explained to EFE by Dr. Raymond Salvador, a biologist and statistician who participated this week in the VII Conference on Schizophrenia held in Palencia.
“We know that skin tissue and nervous tissue have a common origin in the embryo, before differentiating,” explains the researcher.
And therefore, if alterations occur during pregnancy -due to viruses or genetic factors-, it could be expected that these alterations would appear both in the footprint and in the brain.
In addition, it is known that alterations that occur during pregnancy and that alter the development of the fetus increase the risk of suffering from the disease in the future.
Therefore, if fingerprint patterns are fixed before birth – they are trait biomarkers that remain stable throughout life – the fingerprints can be used in individuals from risk groups who have not yet developed the disease as tools. for the prediction of its future evolution, details the researcher.
On this basis, they have been feeding an algorithm with countless data obtained from more than 600 people, half of them people with schizophrenia and the other half healthy, and they have done more than 800 controls.
They then carried out a validation study of the tool designed with Artificial Intelligence that worked well. “We obtained 70% correct answers,” explains the researcher at the Hospital Sisters Research Foundation, FIDMAG, and member of the Mental Health Research Consortium, CIBERSAM.
It is not an extremely high figure, but from a biological point of view this predictive value is important because, as he repeats, the traces are fixed from birth and are unalterable, although there are factors during life, in addition to the genetic part, that can increase the risk of mental illness.
Therefore, the developed tool “offers additional information that helps diagnosis although it is not decisive,” insists the biologist and statistician.
That is, it is not a tool for universal use, but rather it is useful in the so-called “target population”, the population with potential risk indices and genetically predisposed, with mild symptoms or with a family history.
“Support in case of doubt, provide evidence in one direction or another,” he maintains.
The process is very simple. It is enough to take the fingerprints and apply the algorithms to make the calculations that in less than 10 minutes offer a percentage that establishes the probability of suffering from schizophrenia.
“Models are made for each finger and each model gives you a probability of risk,” details the researcher, who insists that the result is an aid for diagnosis.
“This tool can help us know if we need to do a closer follow-up and prevent the development of the disease,” says Dr. Raymond Salvador.
This is very important if one takes into account that in psychoses it has been shown that the sooner action is taken, the better the patient’s evolution is and that if action is not taken soon, the patient’s stabilization is more difficult.
In the next phase, they have proposed applying the same algorithm to the diagnosis of bipolar disorder. “To our surprise, although the algorithm was developed for schizophrenia, it also came out positive for bipolar disorder,” he observes.
This has a double reading. On the one hand, the fact that it comes out positive in the algorithm is good because it detects the risk of another psychosis. But the negative part is that it indicates that schizophrenia and bipolar disorder are not distinguished at the beginning of the disease and differentiating them is very important because the treatment at the therapy and pharmacological level is different.
This means that the designed tool is not specific enough for schizophrenia and more tests have to be done, creating a new database with traces of patients with bipolar disorder to generate a new version of the algorithm that allows a differentiated diagnosis of the two psychoses.
The new field work will allow progress in the application of these same techniques in individuals with bipolar disorder, and they are even working on being able to use it in other neurodevelopmental diseases, such as autism.
