AUTHOR=Zhu Feng-lei , Wang Shi-huan , Liu Wen-bo , Zhu Hui-lin , Li Ming , Zou Xiao-bing TITLE=A multimodal machine learning system in early screening for toddlers with autism spectrum disorders based on the response to name JOURNAL=Frontiers in Psychiatry VOLUME=Volume 14 - 2023 YEAR=2023 URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2023.1039293 DOI=10.3389/fpsyt.2023.1039293 ISSN=1664-0640 ABSTRACT=Background Reduced or absence of response to name (RTN) has been widely reported as an early specific indicator for autism spectrum disorder (ASD), while few studies have quantified RTN of toddlers with ASD in an automatic way. The present study aims to apply multi-modal machine learning system (MMLS) in early screening for toddlers with ASD based on RTN. Methods 125 toddlers were totally recruited, including ASD (n=61), developmental delay (DD, n=31) and typical developmental (TD, n=33). Procedures of RTN were respectively performed by evaluator and caregiver. Behavioral data were collected by 8 definitions tripod-mounted cameras and coded by the MMLS. Response score, response time, and response duration time were accurately calculated to evaluate RTN. Results Total accuracy of RTN scores rated by computers was 0.92. In both evaluator and caregiver procedures, toddlers with ASD had significant differences in response score, response time, and response duration time compared to toddlers with DD and TD (all P <0.05). The area under curve (AUC) was 0.81 for computer rated results, and the AUC was 0.91 for human rated results. The accuracy in identification of ASD based on computer and human rated results was respectively 74.8%, 82.9%. There was a significant difference between the AUC of human rated results and computer rated results (Z =2.71, P =0.01). Conclusions MMLS can accurately quantify behaviors in RTN procedures and effectively distinguish toddlers with ASD from non-ASD. This novel system may provide an accurate and low-cost approach to have an early screening and identification for toddlers with ASD. However, Machine Learning is not as accurate as human observer and the detection of a single symptom like RTN is not sufficient enough to detect ASD.