16183
Ka-O-TV: An Eye Gaze Detector for Early Diagnosis of ASD Phenotype
A range of questionnaires have been invented to detect young children with autism spectrum disorder (ASD), but they rely on caregivers' recall and/or non-expert assessment. Furthermore, they are time-consuming and appear psychologically intrusive for some caregivers, making their responses somewhat biased. Objectivity of the responses therefore may be compromised.
A novel technique, eye gaze detector, provides us objective and useful findings that may be associated with early phenotypes of ASD. Unfortunately, the technique has still not been well founded largely because of limited availability of the devices, difficulty in calibration for very young children, and low data retrieval rate. To challenge these limitations, we invented a new eye gaze detector, "Ka-O-TV", with the technical assistance of Prof. Yoshinobu Ebisawa, Shizuoka University, Japan.
Objectives:
1. To investigate data retrieval rate of Ka-O-TV with a sample of consecutive 1:6- to 2-year-old children.
2. To investigate whether Ka-O-TV successfully differentiate those with and without potential diagnosis of ASD.
Methods:
Participants are representative 1:6- to 2-year-old Japanese children who attended at public health checkup sites held every month in Saga-city, Japan. By the time they were examined with Ka-O-TV, none of the participating children has been confirmed or suspected to have a diagnosis of ASD. Ka-O-TV devices used in the present study were manufactured by JVC Kenwood Corporation. Nice features of the device include the fastest- and easiest-ever calibration; it only takes 5 seconds although effortful attention to the calibration screen is not necessary. Ka-O-TV provides a series of visual stimuli, which have been designed to attract young children at this age and to illuminate social eye gaze features of ASD. It takes less than two minutes to complete the whole process of the presentation of the visual stimuli. Assessment of ASD phenotype was conducted with M-CHAT (Robins et al., 2001) for all the children and with ADOS (Lord et al, 2000) as a part of their full clinical assessment after the health checkup.
Results:
We examined 169 children (M:F = 97:72, age: 19-26 month old). Average data retrieval rate of these children was 0.84. Whilst most children achieved the data rate of >0.90, 9 children (5%) showed the rate lower than 0.50.
Two sets of visual stimuli (preferential looking) were shown to be highly correlated with M-CHAT total scores. The two sets also discriminate children with ASD, confirmed clinically and with ADOS in some children, from those without ASD. The combination of the two stimulus sets provided sensitivity of >90% and specificity of >80%.
Conclusions:
Ka-O-TV was shown to retrieve eye gaze data, quickly and successfully, from representative young children. It also successfully differentiated those with and without an increased risk of ASD.