How AI Can Predict Cognitive Development At Birth

A new study conducted by researchers at the University of North Carolina School of Medicine revealed that a child's cognitive development at two years old could already be predicted at birth with 95 percent accuracy.

In the study published online by Neurolmage journal, the researchers made use of the application of artificial intelligence called machine learning techniques, in addition to MRI brain scans, to arrive at accurate predictions regarding cognitive ability.

Predicting cognitive development using AI

More specifically, a deep learning model was trained using cross-validation, a statistical technique which involves sectioning the data into different subsets, training the data on a certain subset, and using the other subset in order to evaluate the performance of the model. It was then used to classify 75 full-term infants as "scoring above or below the median at age 2 using WM connectomes generated from diffusion weighted magnetic resonance images at birth."

Individual cognitive scores were then predicted using the results from the model. Some of the connections that are considered important for classification included those found within the frontal lobe, as well as those between the frontal lobe and other brain areas. Results from the study revealed that a relatively high accuracy of prediction could be achieved with both full term and preterm infants.

Combining AI and medical technology

Artificial intelligence has certainly progressed greatly throughout the years, thanks to the efforts of researchers and scientists who have been continuously trying to improve the technology needed for AI to be successful.

This study is an interesting example of how two different technologies can work together hand-in-hand. In this case, for example, the MRI brain scans were used to peek into the subject's brain, while the AI looked at all the white matter connections in the brain at birth used them to predict cognitive outcomes as well as prevent birth injuries. Without one or the other, an accurate prediction couldn't have been made.

The thing is, such technologies are vital, especially in a world that is slowly starting to become more aware of mental health. As mentioned in the study, cognitive ability is an important predictor of mental health outcomes that is influenced by neurodevelopment, especially that in children. This means that the ability to foresee a child's cognitive development could help doctors prepare and mitigate problems that could arise later on in the child's life.

The importance of predictions

"This prediction could help identify children at risk for poor cognitive development shortly after birth with high accuracy," explained John H. Gilmore, MD, one of the senior authors of the study. Gilmore also serves as the director of the UNC Center for Excellence in Community Mental Health and happens to be a Thad and Alice Eure Distinguished Professor of psychiatry.

"For these children, an early intervention in the first year or so of life - when cognitive development is happening - could help improve outcomes," Gilmore added, citing premature infants at risk as an example. In that case, he said, "one could use imaging to see who could have problems."

Gilmore also mentioned that many researchers, not just at UNC, are working hard to search for imaging biomarkers that they can use for weak cognitive outcomes, as well as for risk of various neuropsychiatric conditions that could surface later on in life.

Understanding neuropsychiatric conditions

Neuropsychiatric conditions fall under eight main categories: addiction, such as alcohol and drug dependence; childhood development disorders, such as hyperactivity disorder (ADHD), and autism; eating disorders, such as bulimia nervosa and anorexia nervosa; degenerative disorders, such as Alzheimer's and dementia; mood disorders, such as depression and bipolar disorder; neurotic disorders, such as obsessive compulsive disorder (OCD) and anxiety disorder; psychosis, such as schizophrenia; and sleep disorders, such as insomnia and sleep apnea.

As of last year, neuropsychiatric disorders are the newest leading cause of disability in the U.S., followed by cardiovascular and circulatory diseases and neoplasms, according to the National Institute of Mental Health.

In fact, they contribute to almost 18.7 of U.S. DALYs, or disability-adjusted life years. This number represents the total number of years lost to disability, illness, or premature death within a given population, and it can be calculated by adding the number of years of life lost to the number of years lived with disability for a certain disorder or disease.

Artificial intelligence for mental health awareness

If we can develop artificial intelligence to be able to predict a two-year child's cognitive development right at birth, we may also be able to predict other neuropsychiatric conditions and hopefully prevent them from affecting the population. At the very least, they could be used to help reduce the risk of developing such conditions.

"Our study finds that the white matter network at birth is highly predictive and may be a useful imaging biomarker," Gilmore continues. "The fact that we could replicate the findings in a second set of children provides strong evidence that this may be a real and generalizable finding."