Initiative by Unifesp researchers had the support of FAPESP
Researchers at the Escola Paulista de Medicina (EPM) of the Federal University of São Paulo (Unifesp) have developed a software that detects facial expressions related to the sensation of pain in newborns, which may help in the care of babies, enabling more agile and precise interventions .
The computer program was conceived within the scope of the Software development research to identify the facial expression of pain in newborns, conducted by Ruth Guinsburg with support from FAPESP.
According to the researcher, the initiative arose from the difficulty faced by caregivers of newborns in intensive care units (ICU) in recognizing and evaluating the signs of pain. “This subjectivity ends up making possible interventions difficult, since there are a number of factors that can lead the newborn to demonstrate certain discomforts that are not always related to pain. The research provides a useful tool to monitor the baby's pain in the routine of neonatal units”.
Pain recognition is based on behavioral and physiological indicators, such as simple motor responses, facial expressions and crying. The software was developed based on the Neonatal Facial Coding System (NFCS), widely used in the recognition of facial movements of pain, converted by researchers into computer language with the collaboration of the Department of Health Informatics (DIS) of EPM and professionals from the University of Mogi das Cruzes.
The software mechanisms began to be conceived in 2009, after the project was approved by the Research Ethics Committee at Unifesp. 30 newborns were filmed at Hospital São Paulo, between June and August 2013. “It was necessary to work very carefully with the families so that there were no misunderstandings about the capture of images, performed during painful procedures with medical indication, such as capillary puncture , venous or arterial and intramuscular or subcutaneous injection, necessary for the care of newborns and not carried out for research purposes”, explains Ruth.
The software, based on biometric identification, mapped and detected 66 points on the babies' face, then reduced to 16 main nodal points, from which those who moved the most when acute pain caused by some medical procedure was expressed.
The distances between the points were the basis for detecting facial expressions that, according to the scale adopted by the research, show signs of pain: protruding forehead, narrowed palpebral fissure, deepened nasolabial fold, open mouth and tense mouth.
5,644 images were identified, an average of 188 per newborn. Next, the researchers tested the agreement between the software's analyzes and those of six health professionals experienced in the recognition of neonatal pain, specializing in neonatology. Three images of each baby were compared: two recorded at rest, without pain, and one during painful procedures.
The precision with which the software detected expressions makes it an important tool to help healthcare teams in neonatal ICUs. “The newborn's pain is always seen by another and depends on the caregiver's decoding, subject to the time available for monitoring and even more subjective factors, such as empathy. The automation of this follow-up can contribute to the babies' well-being and help with care, leading to more assertive interventions”, says Ruth.
The researchers are now working on adapting the software for monitoring sick babies, improving the cameras and adapting the system so that it can be used at the bedside.
Source: FAPESP Agency
The computer program was conceived within the scope of the Software development research to identify the facial expression of pain in newborns, conducted by Ruth Guinsburg with support from FAPESP.
According to the researcher, the initiative arose from the difficulty faced by caregivers of newborns in intensive care units (ICU) in recognizing and evaluating the signs of pain. “This subjectivity ends up making possible interventions difficult, since there are a number of factors that can lead the newborn to demonstrate certain discomforts that are not always related to pain. The research provides a useful tool to monitor the baby's pain in the routine of neonatal units”.
Pain recognition is based on behavioral and physiological indicators, such as simple motor responses, facial expressions and crying. The software was developed based on the Neonatal Facial Coding System (NFCS), widely used in the recognition of facial movements of pain, converted by researchers into computer language with the collaboration of the Department of Health Informatics (DIS) of EPM and professionals from the University of Mogi das Cruzes.
The software mechanisms began to be conceived in 2009, after the project was approved by the Research Ethics Committee at Unifesp. 30 newborns were filmed at Hospital São Paulo, between June and August 2013. “It was necessary to work very carefully with the families so that there were no misunderstandings about the capture of images, performed during painful procedures with medical indication, such as capillary puncture , venous or arterial and intramuscular or subcutaneous injection, necessary for the care of newborns and not carried out for research purposes”, explains Ruth.
The software, based on biometric identification, mapped and detected 66 points on the babies' face, then reduced to 16 main nodal points, from which those who moved the most when acute pain caused by some medical procedure was expressed.
The distances between the points were the basis for detecting facial expressions that, according to the scale adopted by the research, show signs of pain: protruding forehead, narrowed palpebral fissure, deepened nasolabial fold, open mouth and tense mouth.
5,644 images were identified, an average of 188 per newborn. Next, the researchers tested the agreement between the software's analyzes and those of six health professionals experienced in the recognition of neonatal pain, specializing in neonatology. Three images of each baby were compared: two recorded at rest, without pain, and one during painful procedures.
The precision with which the software detected expressions makes it an important tool to help healthcare teams in neonatal ICUs. “The newborn's pain is always seen by another and depends on the caregiver's decoding, subject to the time available for monitoring and even more subjective factors, such as empathy. The automation of this follow-up can contribute to the babies' well-being and help with care, leading to more assertive interventions”, says Ruth.
The researchers are now working on adapting the software for monitoring sick babies, improving the cameras and adapting the system so that it can be used at the bedside.
Source: FAPESP Agency
EN 
PT 
