Analyzing babies’ vocalizations using machine learning could aid in early diagnosis of Rett syndrome, a small study has found.
Researchers say such voice analysis in infants aged 6 to 11 months can help identify signs of Rett, or fragile X syndromelong before a child is typically diagnosed with these disorders in early childhood.
“We…demonstrate that, although individuals studied with [fragile X] and [Rett] had not yet shown clinical signs during their second half of life, the machine “hears” that they had already vocalized differently from TD [typically developing] individuals,” the researchers wrote.
The team noted that these early results were based on data from a small number of individuals and stressed the need for further research to confirm and extend this approach.
The study, “Automatic vocalization-based detection of Fragile X Syndrome and Rett Syndromewas published in the journal Scientific reports.
Both Rett and Fragile X are developmental disorders with symptoms usually beginning in early childhood. Speech abnormalities are usually associated with these two genetic diseases.
In Rett syndrome, as well as fragile X syndrome, patients often experience delays in getting a correct diagnosis. Many symptoms are common to a range of disorders and are not enough on their own to diagnose the conditions.
Survey of vocalizations for the diagnosis of Rett syndrome
Now, a team of scientists in Germany and Austria have tested the idea that automated analyzes of voice recordings could be useful for the early identification of Rett and Fragile X.
The study included three babies with a diagnosis of Rett syndrome and three others with Fragile X. The infants were 6 to 11 months old.
All infants with Rett were female, consistent with typical gender-based prevalence of the disorder, while all infants with Fragile X were male. A set of six typically developing babies, matched for age and sex, were included as controls.
Very simply, the analysis involved taking audio data from home recordings of children and then feeding the data into a computer with a set of mathematical rules. The computer would then use these rules to “learn” how to sort the children into predefined groups.
In a first series of tests, researchers assessed whether these scans could differentiate children with Rett or Fragile X from sex-matched controls, or whether they could differentiate between abnormal development (Rett or Fragile X) and development typical. In these first analyses, the computer worked with 100% accuracy.
These results “not only demonstrate basic feasibility, but underscore the approach’s high potential for future practical application in pediatric healthcare,” the scientists wrote.
Additionally, closer examination of the data showed that the voice features that were important in making these differentiations in the computational models were distinct for Rett compared to Fragile X.
“This suggests that the early verbal idiosyncrasies of individuals with [fragile X] and people with [Rett] manifest acoustically in different ways compared to the typical early verbal behavior of same-sex controls,” the researchers wrote.
In other analyses, investigators tried to use voice-based analysis to sort all children into the appropriate group – Rett, Fragile X, or Control.
“The present study was the first-ever attempt to combine early vocalization data from individuals with different late-detected genetic disorders into a classification model,” the team noted.
Of the 12 children, nine were correctly classified. One child with Fragile X was incorrectly classified as having Rett, while another child with Fragile X and one with Rett were incorrectly classified as having typically developed.
The researchers outlined some potential changes to the algorithm that could be helpful in improving accuracy. They also stressed the importance of further research with larger datasets to validate and refine the approach.
“Although our results indicate that this approach is [worthwhile] to be followed further, they must be interpreted with caution and can hardly be generalized. This is mainly due to the very small dataset,” the scientists wrote.
The fact that the analyzes are based on home video recordings was also noted as a limitation.
“It should be borne in mind that the recordings were not originally made with the intention of the parents to collect data for later scientific analysis, but generally to create a memory of family routines and special moments. of their children’s early childhood,” the researchers wrote. noting that this may lead to an under-representation of abnormal behaviors that parents chose not to record.
Despite these limitations, the researchers said the use of home video “offers the unique chance to study early development in a natural setting and currently represents the best available approach to objectively investigate prodromal disease.” [early] behavioral phenomena in rare late-detected developmental disorders such as [fragile X] Where [Rett].”
