Detecting Problem Behavior in Children from Biological Signals: Validation Through Chaos Analysis

Tsujino, Junko; Oyama-Higa, Mayumi; Tanabiki, Mitsuko

The relationship between biological signals and problem behavior in children was examined in a total of 87 children from a class of two-year-olds and a class of three-year-olds, with input from their mothers. The mothers were asked to fill out a checklist on the behavior of their children. Fingertip pulse waves were measured for use as biological data. The measured values were subjected to chaos analysis and the Lyapunov exponents calculated. The two-year-olds and the three-year-olds differed in the scores for "Withdrawn" and "Developmental Abnormality" as assessed by the criteria used in the behavior checklist. The two classes also differed significantly in their Lyapunov exponents. Among the two-year-olds, the Lyapunov exponent differed significantly between those who were withdrawn and those who were not. Among the three-year-olds, there was a significant difference in the Lyapunov exponent between children with severe attention problems and those with only minor attention problems. The Lyapunov exponent calculated by chaos analysis of fingertip pulse wave data appeared to be an effective index for scientifically studying problem behavior in young children. © 2006 IEEE.