AI is used to detect emotions and weigh cattle – 08/21/2023 – Tech

The use of Artificial Intelligence (AI) in work and research has become part of the daily life of public universities in the country. They are aimed at the most diverse areas, such as agro, health and astronomy. Some have the potential to directly influence the lives of the population in the coming years.

One of the initiatives is the work that is being developed at UFMG (Federal University of Minas Gerais), through the doctorate of production engineer Daniela Castro Araújo. With the use of AI, she is working on a study in which it will be possible to detect early breast cancer through a blood test.

“We had access to exams of thousands of women with prior knowledge about whether or not they had breast cancer. We used Artificial Intelligence to identify patterns and differences. Once trained, this model can be used to predict any future blood test”, he says Araújo.

The model does not replace mammography, according to the researcher. “Mammography is the gold standard for screening. The only problem is that it is not accessible to everyone. What our tool does is a personalized risk stratification. Looking at the blood test, we can tell if you are in a higher risk range, in an intermediate risk range or in a lower risk range. Those women who are in the highest risk range should be prioritized for a monograph.”

The tool has entered what researchers call the “real world”. Tests are being carried out with some health operators looking for women who are already in arrears with the mammography exam, according to the researcher. There is no word on when the service will be available to the public.

In the agro area, a study at UFMT (Federal University of Mato Grosso), in partnership with IFMT (Federal Institute of Education, Science and Technology of Mato Grosso) and other institutions, is developing a tool in which the farmer will be able to weighing cattle using a photograph taken with a cell phone.

Mathematical models were created from morphometric measurements, with the idea of ​​predicting the information on the animal’s carcass.

“From the point of view of animal production, it is necessary to have weight information to formulate a diet, to know if the animal is ready for slaughter, to apply the medicine”, says Geovanne Ferreira Rebouças, doctor in zootechnics and professor at IFMT.

“The large properties already have scales, but the small and medium producers do not, which makes this more technical information difficult.”

The research is still in the stage of collecting images of the animals. The trend is that in the second half of 2024 the application will enter the testing phase for validation.

Emotion detector can help patients

Another study has sought to identify, from the electrical signals generated by the brain, which emotion the person was feeling, with a classification that includes, for example, sadness, joy, astonishment and pain. It is being developed at UFLA (Federal University of Lavras).

The work began as research for the master’s thesis by engineer Vancley Oliveira Simão. The researchers trained some Artificial Intelligence methods so that the system could identify which emotion the person was feeling.

They used an existing database in a laboratory in the United States. The signals are captured through a cap placed on the person’s head. According to William Soares Lacerda, a professor at the Department of Automatics at the UFLA School of Engineering, the accuracy was approximately 90%.

“This work can be applied, for example, to a person who is bedridden, unable to communicate. Her brain continues to function perfectly, but she is unable to speak, unable to express what she is feeling”, says Lacerda.

At Unesp, research focused on astronomy has already earned researchers an award. They developed a methodology capable of reducing the analysis of an asteroid, which could take months, to just a few seconds.

With the use of AI, three different artificial neural networks were tested to analyze the dynamics of asteroids. With this work, it was possible to reduce the analysis time and also allow the identification of young families of asteroids.

“In the past, it was a job of visual observation, with hundreds of asteroid charts. Now, with the volume of data, that would be unfeasible. We works with more than 1.2 million asteroids”, explains Valerio Carruba, professor of the mathematics department at Unesp, about the need to use AI.

“Now, not only do we have a way of knowing the age of these objects, which is three million years old, but we can also calculate the initial speed with which they were ejected. This information is very rare because, with older families, it is very difficult to be obtained”, completes the researcher.

This work won the prize for “Innovative computational methods in astronomical dynamics” at the 8th International Meeting of Celestial Mechanics. The award ceremony will take place in September, in Italy.

“The concern we have every time an article is published is to make the code available so that it has more applications, so that this network grows, not only in the area of ​​astronomy”, says Rita de Cássia Domingos, a researcher at the Faculty of Engineering from Unesp.

Despite advances in technology, researchers say there is a lack of people prepared to work with Artificial Intelligence. “Researchers are lacking the ability to use these tools and the ability to stay current in a rapidly evolving area,” says Carruba.

“Who is working, is learning. Those who really dominate are IT technicians, as it is something more natural for them. However, it is necessary to form multidisciplinary teams to remedy this lack”, says Geovanne Ferreira Rebouças, from IFMT.