Algorithm tells us why the baby cries
American researchers develop an algorithm to understand the crying of the newborn and to distinguish normal crying from the one who has to worry
An algorithm to understand the crying of the newborn
The crying is the primary means of communication for children and every parent knows the frustration of responding to the cries of a child, wondering why she cries incessantly? Is he hungry, wet, tired, in need of a hug or maybe he feels pain? With time and experience, you learn to recognize most types of crying, but for beginners and inexperienced parents, things can be really difficult.
A new artificial intelligence algorithm , based on automatic speech recognition, would be able to accurately identify characteristics in a baby’s crying and to distinguish between normal and abnormal crying signals, such as those arising from a disease . The research was conducted by scientists at Northern Illinois University , under the guidance of Lichuan Liu, and was published in the May issue of IEEE / CAA Journal of Automatica Sinica (JAS).
Why is the baby crying?
The new research uses a specific algorithm based on automatic speech recognition to detect and recognize the characteristics of infantile crying .
The authors examined some newborns in a neonatal intensive care unit and analyzed their crying using an artificial intelligence algorithm resulting from a complex voice recognition system. To create it, scientists analyzed and classified the various types of crying using a technique called ” compressed sensin g” which allows them to be acquired more efficiently even in noisy environments .
In a second phase the scientists asked neonatal nurses to provide the meaning of some types of crying detected. For example:
- The ” neh ” indicates in most cases that the child is hungry and is produced when the tongue is pushed on the palate and the newborn has the reflection of sucking,
- an acute ” heh ” indicates other needs and is produced by the child when he is cold, needs to be changed or wants to change position,
- the sound ” owh ” is generally linked to the reflection of a yawn and indicates that the child is sleepy,
- ” eair ” is a deeper sound that comes from the abdomen, so it probably means that the child has pain in the abdomen, if it is a colic this sound is accompanied by typical behaviors such as pulling the knees upwards.
- At this point, scholars have combined both the data collected by the algorithm and those provided by the neonatal nurses to be able to make a classification of the types of crying related to the different emotions that cause them. Two large groups of tears have emerged :
- the normal crying , which includes the first four sensations (hunger, sleep, need to be changed and to receive care),
- the abnormal crying, caused by a nuisance and which can also be associated with the presence of a passenger disturbance or illness.
- Because this algorithm can be important To date, studies conducted in this regard have analyzed only normal crying signals – hunger, a wet diaper and the need for attention – while no research had focused on the possibility of being able to interpret the crying related to more important conditions that require a intervention. Abnormal crying is often related to medical problems such as:
- disorders of the central nervous system,
- hypopharynx trauma,
- vocal cord paralysis
- many other pathological and chronic conditions. The researchers explain that they have created, therefore, a new algorithm for recognizing crying language, able to distinguish normal crying signals from abnormal ones in a noisy and uncontrolled environment (as in real everyday life).
In other words, compared to normal crying signals, the anomalous cry signals are more intense and require further evaluation, hence the importance of having a tool that can help make them more recognizable.
The extraordinary peculiarity of this algorithm is that it can represent a real non-invasive diagnosis of the child’s health conditions without the use of invasive tests and could be used in different situations such as:
- to address medical problems for which there is currently no diagnostic tool available, such as the Sudden Infant Death Syndrome (SIDS), developmental and colic problems ,
- face medical problems that can only be diagnosed by invasive procedures, such as chromosomal abnormalities ,
- to address medical problems that may have already been identified but that can be managed better with a tool that can improve the definition of prognosis, for example prematurity .