Scientists have trained neural networks to successfully detect DNA damage

Neural networks effectively predict changes in DNA structure caused by UV radiation.

machine learning dna analysis picture
machine learning dna analysis

Scientists at Tomsk Polytechnic University and the University of Chemistry and Technology of Prague conducted a series of experiments that proved that artificial neural networks can be successfully used in medical diagnostics to detect DNA damage from ultraviolet radiation. 

Thanks to the development of TPU researchers, the sensor can detect toxic substances, heavy metals, as well as some diseases and defects in the DNA structure. 

The advantages of "hybrid" sensors are hypersensitivity, speed of analysis and the possibility of analysis at the sampling time.

According to the authors of the article, at present, it remains practically unexplored how exactly ultraviolet radiation damages the DNA structure, especially with short-term exposure. 

It is known that the effect of ultraviolet radiation can cause cancer. However, minor changes in the structure of DNA itself by classical methods are almost impossible to detect.

The use of neural networks allowed scientists to avoid the mathematical processing of a huge number of spectra and the need to optimize the measurement procedure itself. 

In addition, the neural networks not only reveal the damage themselves, but also very effectively predict changes in the DNA structure caused by UV radiation.

Researchers believe that neural networks and spectroscopy in the future can be successfully used for medical diagnostics. Moreover, this technique can be further improved.

Machine Learning genomics data chart
Machine Learning genomics

Damage to DNA under the influence of UV radiation was an extremely interesting model to study. This concept, which allows detecting minimal changes in the structure of DNA, can be expanded and improved.