Automated AI based identification of autism spectrum disorder from home videos - Scorecard - MDSpire

Automated AI based identification of autism spectrum disorder from home videos

  • By

  • Dong Yeong Kim

  • Ryemi Do

  • Youmin Shin

  • Hewoen Sim

  • Hanna Kim

  • Sungchul Cho

  • Geonhee Lee

  • Seyeon Park

  • Boa Jang

  • Hyojeong Lim

  • Sungji Ha

  • Jaeeun Yu

  • Hangnyoung Choi

  • Junghan Lee

  • Min-Hyeon Park

  • Ayeong Cho

  • Chan-Mo Yang

  • Dongho Lee

  • Heejeong Yoo

  • Yoojeong Lee

  • Guiyoung Bong

  • Johanna Inhyang Kim

  • Haneul Sung

  • Hyo-Won Kim

  • Eunji Jung

  • Seungwon Chung

  • Jung-Woo Son

  • Jae Hyun Yoo

  • Sekye Jeon

  • Jinseong Jang

  • You Bin Lim

  • Jeeyoung Chun

  • Wooseok Choi

  • Sooyeon Lee

  • Sohyun Park

  • Jisung Ahn

  • Chae Rim Lee

  • Keun-Ah Cheon

  • Young-Gon Kim

  • Bung-Nyun Kim

  • October 10, 2025

  • 0 min

Share

Clinical Scorecard: AI-Driven Detection of Autism Spectrum Disorder Utilizing Home Video Analysis

At a Glance

CategoryDetail
ConditionAutism Spectrum Disorder (ASD)
Key MechanismsAutomated analysis of short home-recorded videos using AI and machine learning to extract behavioral features for early ASD detection
Target PopulationChildren aged 18–48 months
Care SettingHome environment with remote video-based screening; complements clinical evaluation in hospitals and resource-limited settings

Key Highlights

  • Early ASD diagnosis is critical but often delayed due to resource-intensive traditional assessments and systemic barriers.
  • AI-based screening using brief home video protocols (name-response, imitation, ball-playing) achieved 0.83 AUC and 0.75 accuracy in distinguishing ASD from typical development.
  • Home videos capture naturalistic child behaviors with high ecological validity, addressing limitations of clinical and caregiver-report assessments.

Guideline-Based Recommendations

Diagnosis

  • Incorporate AI-driven home video analysis as a complementary screening tool to prioritize referrals for formal ASD diagnostic evaluation.
  • Use brief, standardized home video protocols to elicit natural behaviors relevant to ASD detection.

Management

  • Enable earlier intervention by facilitating timely identification through scalable, automated screening methods.
  • Integrate AI screening results with clinical assessments to guide personalized intervention planning.

Monitoring & Follow-up

  • Utilize repeated home video screenings to monitor behavioral changes over time in naturalistic settings.

Risks

  • Be aware of potential limitations in AI model generalizability due to sample size and variability in home video quality.
  • Avoid over-reliance on caregiver-report tools alone due to limited specificity and sensitivity.

Patient & Prescribing Data

Children aged 18–48 months undergoing ASD screening

AI-based home video screening can identify children at risk earlier than traditional methods, enabling timely referral and intervention.

Clinical Best Practices

  • Employ short, task-based home video protocols under 1 minute each to capture naturalistic behaviors.
  • Combine AI-extracted behavioral features with demographic data for improved screening accuracy.
  • Use automated analysis to reduce observer bias and increase scalability in diverse clinical and community settings.

References

Original Source(s)

Automated AI based identification of autism spectrum disorder from home videos

  • by Dong Yeong Kim, Ryemi Do, Youmin Shin, Hewoen Sim, Hanna Kim, Sungchul Cho, Geonhee Lee, Seyeon Park, Boa Jang, Hyojeong Lim, Sungji Ha, Jaeeun Yu, Hangnyoung Choi, Junghan Lee, Min-Hyeon Park, Ayeong Cho, Chan-Mo Yang, Dongho Lee, Heejeong Yoo, Yoojeong Lee, Guiyoung Bong, Johanna Inhyang Kim, Haneul Sung, Hyo-Won Kim, Eunji Jung, Seungwon Chung, Jung-Woo Son, Jae Hyun Yoo, Sekye Jeon, Jinseong Jang, You Bin Lim, Jeeyoung Chun, Wooseok Choi, Sooyeon Lee, Sohyun Park, Jisung Ahn, Chae Rim Lee, Keun-Ah Cheon, Young-Gon Kim, Bung-Nyun Kim

  • October 10, 2025

Related Content

Invisible Disabilities in Medical Education: The Oversights and Their Impact on Patients

Exploiting large language models for diagnosing autism associated language disorders and identifying distinct features

Casey Means, a MAHA leader, makes bid to be nation’s top doctor