Skip to content
DraconiticResearch
Search Sources API
182,549 studies31 sourcesUpdated 30 min ago
Peer-reviewedCC BY-NC-NDNature communications

Automated screening of autism via a pretrained vision-language model in naturalistic caregiver-child interactions.

  • Ruikang Deng1,2
  • Huishi Huang3
  • Shaoli Lv3
  • Yu Xing3,4
  • Yijie Li3
  • Cong You3
  • Fei Wu1
  • Hongzhu Deng5,6
  • Zhengxing Huang7
7 affiliations
  1. 1College of Computer Science and Technology, Zhejiang University, Hangzhou, China.
  2. 2School of Medicine, Shanghai Jiao Tong University, Shanghai, China.
  3. 3Child Developmental and Behavioral Center, Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China.
  4. 4Child Developmental and Behavioral Center, Zhaoqing Hospital of the Third Affiliated Hospital of Sun Yat-sen University, Zhaoqing, China.
  5. 5Child Developmental and Behavioral Center, Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. [email protected].
  6. 6Child Developmental and Behavioral Center, Zhaoqing Hospital of the Third Affiliated Hospital of Sun Yat-sen University, Zhaoqing, China. [email protected].
  7. 7College of Computer Science and Technology, Zhejiang University, Hangzhou, China. [email protected].
Read full textPDFDOICite

Abstract

Early autism screening is essential for timely intervention but remains constrained by specialist dependence, costly protocols, and limited accessibility. We present Autism-CLIP, a vision-language model that automates screening from 3-min naturalistic caregiver-child interaction videos recorded in toddlers aged 15-24 months. By integrating contrastive learning, large language model-generated supervision, domain-specific architecture, and hybrid training, Autism-CLIP aligns visual behavioral cues with autism-related assessment outcomes, enabling annotation-free inference. In multicenter evaluation, Autism-CLIP achieved an AUC of 0.903 in the internal cohort (n = 103) and 0.873 in the external cohort (n = 102), outperforming conventional machine learning methods. Attention mapping highlighted clinically relevant markers, including gaze aversion and limited social reciprocity. Requiring only smartphone-captured videos without manual coding or controlled settings, Autism-CLIP offers a scalable, low-cost approach to broaden access to early autism screening and support prioritization of clinical resources.

Related

  1. CytosMPS: an automated multimodal pooled screening platform for protein, morphology and transcriptomics
    T. Lopez, J. Rabalais, B. Song +6
    bioRxivPreprintCC BY-NCFull textCell biologybioRxiv
  2. Genotype × Isolate Interactions for Tan Spot Resistance Screening in Mungbean
    Krishna Chaitanya Alamuru, Noel L. Knight, Prue Bottomley +3
    Plants (Basel, Switzerland)Peer-reviewedCC BYFull textPlant biologyEurope PMC
  3. Optimizing plant species selection for automated monitoring of plant-pollinator interactions
    Y. Zhong, J. B. Lanuza, J. M. Heuschele +3
    bioRxivPreprintCC BYFull textEcologybioRxiv
  4. Automated screening of Parkinson's Disease using Smooth Pursuit Eye Movements
    J. Camilo Puerta-Acevedo, Maria F. Alcala-Durand, Julián D. Arias-Londoño +3
    arXivPreprintCC BY 4.0Full textBioinformatics & computational biologyarXiv
  5. Interactions between automated cooling irrigation and basal leaf removal in cv. Barbera grapevines
    Paolo Bonini, Federico Terzi, Harsh Tiwari +1
    Scientific ReportsPeer-reviewedCC BYFull textPlant biologyDOAJ

Cite

BibTeXRIS
BibTeX
@article{deng2026automated,
  title = {{Automated screening of autism via a pretrained vision-language model in naturalistic caregiver-child interactions.}},
  author = {Ruikang Deng and Huishi Huang and Shaoli Lv and Yu Xing and Yijie Li and Cong You and Fei Wu and Hongzhu Deng and Zhengxing Huang},
  journal = {Nature communications},
  year = {2026},
  month = sep,
  doi = {10.1038/s41467-026-77721-8},
  url = {https://europepmc.org/article/MED/42893978},
  note = {CC BY-NC-ND},
}
RIS
TY  - JOUR
TI  - Automated screening of autism via a pretrained vision-language model in naturalistic caregiver-child interactions.
AU  - Ruikang Deng
AU  - Huishi Huang
AU  - Shaoli Lv
AU  - Yu Xing
AU  - Yijie Li
AU  - Cong You
AU  - Fei Wu
AU  - Hongzhu Deng
AU  - Zhengxing Huang
PY  - 2026
DA  - 2026/09/16
JO  - Nature communications
DO  - 10.1038/s41467-026-77721-8
UR  - https://europepmc.org/article/MED/42893978
AB  - Early autism screening is essential for timely intervention but remains constrained by specialist dependence, costly protocols, and limited accessibility. We present Autism-CLIP, a vision-language model that automates screening from 3-min naturalistic caregiver-child interaction videos recorded in toddlers aged 15-24 months. By integrating contrastive learning, large language model-generated supervision, domain-specific architecture, and hybrid training, Autism-CLIP aligns visual behavioral cues with autism-related assessment outcomes, enabling annotation-free inference. In multicenter evaluation, Autism-CLIP achieved an AUC of 0.903 in the internal cohort (n = 103) and 0.873 in the external cohort (n = 102), outperforming conventional machine learning methods. Attention mapping highlighted clinically relevant markers, including gaze aversion and limited social reciprocity. Requiring only smartphone-captured videos without manual coding or controlled settings, Autism-CLIP offers a scalable, low-cost approach to broaden access to early autism screening and support prioritization of clinical resources.
SN  - 2041-1723
ER  -