Child in smart jumpsuit (Sampsa Vanhatalo, University of Helsinki)
14 Feb. 2020. A one-piece garment fitted with sensors is shown to capture data about an infant’s movements for analysis by algorithms to reveal possible developmental problems. A description of the baby’s high-tech play wear and the algorithms for analysis appear last month in the journal Scientific Reports.
Researchers from University of Helsinki and Aalto University in Espoo, Finland are seeking more objective methods for assessing a child’s early neurological development, often revealed by atypical movements of the limbs and torso. While developmental conditions such as cerebral palsy and autism spectrum disorders can be detected early on by observing infants at play, these observations can be subjective and are not easily quantified. In addition, children may not spontaneously play in an unfamiliar setting, like a doctor’s office, thus capturing a child’s movements unobtrusively while at home would likely provide more accurate data.
A team led by University of Helsinki neuroscientist Sampsa Vanhatalo developed a one-piece garment resembling a jumpsuit for infants as young as five months to wear while they play at home. Sewn into the soft fabric are accelerometer and gyroscope sensors devised by Movesense, an open-source wearable project, with the devices positioned in the upper arms and legs. Data captured by the sensors are sent via Bluetooth to a nearby receiver. There, the data are processed and displayed with software by the German company Kaasa, a participant in the Movesense project.
Vanhatalo and colleagues asked parents of 22 typically-developed infants about seven months old, to have their children wear the jumpsuits at home. While the infants played in the jumpsuits, the devices captured data sent to the receiving systems, and were also video-recorded. Child development experts then observed the videos of infants at play and annotated the clips with a characterization scheme for posture and movement patterns of children that age.
With that movement characterization scheme, the researchers developed algorithms for analyzing movement data collected by sensors in infants’ jumpsuits. The algorithms use a convolutional neural network, a type of artificial intelligence that combines image analysis and deep machine learning to dissect an image by layers for understanding features in the image. Different aspects of each layer discovered and analyzed by the system are translated into data that the algorithm then uses to train its understanding of the problem being solved, with that understanding enhanced and refined as more images and data are encountered.
The team carried out a series of experiments comparing expert observations of the children’s motility or independent movements to results produced by algorithms from the sensor data. One proof-of-concept test evaluated results of experts’ ratings of five children with higher motor performance and five lower-performing infants. Separate assessments using data from the jumpsuit sensors and analyzed by the algorithm returned similar results.
“The smart jumpsuit provides us with the first opportunity to quantify infants’ spontaneous and voluntary movements outside the laboratory,” says Vanhatalo in a Helsinki University statement. He adds, “The measurements provide a tool to detect the precise variation in motility from the age of five months, something which medical smart clothes have not been able to do until now.”
More from Science & Enterprise:
- Wearable Sensor Analyzes Sweat, Measures Chemicals
- Graphene Material Designed for Wearable Devices
- Smart Fabrics Boost Wearable Sensor Connections
- FDA Clears Wearable App Device for Migraine
- Wearable Patch Designed for Personal Warmth, Cooling
* * *
RSS - Posts
You must be logged in to post a comment.