Application of Telemedicine for Preliminary Screening of Autism Spectrum Disorder

Ting Qiu, Heng Zhang, Conghua Zhou, Qilong Tang, Lizhen Wang, Xiaoyan KeFrontiers in Pediatrics 2022 · DOI

This research paper, published in Frontiers in Pediatrics, explores the transition of Autism Spectrum Disorder (ASD) screening from traditional, labor-intensive offline methods to a streamlined, mobile-first telemedicine approach. By leveraging the ubiquitous WeChat platform, researchers in China developed a system that maintains high diagnostic accuracy while significantly lowering the barriers to early detection.

TL;DR

Researchers developed and validated a WeChat-based telemedicine system for early ASD screening in toddlers, achieving a sensitivity of 92% and specificity of 90%. This digital approach addresses the "bottleneck" of traditional offline screening—high costs and a shortage of specialized staff—making large-scale population screening feasible in resource-limited settings.

What Problem Does This Paper Address?

Early diagnosis and intervention are the "gold standards" for improving the long-term prognosis of children with ASD. However, in many regions, particularly in China, screening is a manual, offline process.

The authors identify several critical pain points:

  • Resource Scarcity: There is a significant shortage of specialized pediatricians and psychologists to conduct screenings.
  • Infrastructural Limits: Community health centers often lack the private, quiet rooms required for standardized behavioral observations.
  • Logistical Friction: Paper-based questionnaires are costly to print, slow to process, and prone to human error during manual scoring.
  • The "Caregiver Gap": In China, grandparents often take children to health checkups but may lack the specific behavioral knowledge required to answer screening questions accurately. A mobile system allows primary caregivers (parents) to provide data remotely.

Methods

The researchers built an online screening system using a Browser/Server (B/S) architecture integrated into the WeChat public platform.

  • The Tool: They used the CHAT-23-A, the Chinese-validated version of the Checklist for Autism in Toddlers. This 23-item parent-report questionnaire is a standard for preliminary screening.
  • The Tech: Developed using PHP and MySQL, the system automates the entire workflow: a parent scans a QR code, completes the digital form, and the system instantly calculates the score based on established clinical criteria (e.g., a positive screen is defined as failing $\ge$ 6 of 23 items or $\ge$ 2 of 7 "key" items).
  • Communication Loop: If a child screens positive, the system automatically alerts both the parent and a doctor, facilitating an immediate transition to online consultation or a follow-up appointment.

Datasets

The study involved a cohort of 100 children aged 18–30 months:

  • ASD Group (n=50): Children previously diagnosed with ASD using DSM-5 criteria and the Childhood Autism Rating Scale (CARS).
  • Control Group (n=50): Typically developing (TD) children, sex-matched to the ASD group.
  • Exclusion Criteria: Children with systemic diseases, head injuries, or other neurological/psychiatric illnesses were excluded to ensure the tool's specificity to ASD was being tested.

Key Insights and Ideas

  • Democratizing Diagnostics: The most significant "aha moment" is the shift from a passive healthcare model (waiting for parents to visit a specialist) to an active model (screening children in their own communities via smartphone).
  • High Diagnostic Fidelity: Despite moving to a digital format, the system's Area Under the Curve (AUC) was 0.934, indicating "excellent" performance that rivals traditional face-to-face screening.
  • Closing the Feedback Loop: Unlike paper forms that might sit in a file, the digital system provides "timely feedback." The automated dialog box that prompts parents of high-risk children to seek further screening is a crucial nudge in the clinical pathway.
  • Data Aggregation: The internal management system allows health authorities to aggregate screening data across different community centers, providing a real-time "heat map" of ASD prevalence and screening rates.

Key Results

The quantitative results suggest the telemedicine system is highly reliable:

  • Sensitivity: 0.92 (The system correctly identified 46 out of 50 ASD cases).
  • Specificity: 0.90 (The system correctly identified 45 out of 50 typically developing children).
  • Accuracy Metrics: The positive predictive value was 0.902, and the negative predictive value was 0.918.
  • Kappa Value: 0.802, representing strong agreement between the digital tool and gold-standard clinical diagnoses.

Limitations and Future Work

While promising, the authors acknowledge several hurdles:

  • Selection Bias: The study used children who were already diagnosed or highly suspected of having ASD. The system's performance might vary when applied to a general, "blind" population where symptoms are more subtle.
  • Sample Size: A 100-person cohort is relatively small for validating a tool intended for "large-scale population screening."
  • Differential Diagnosis: Future research needs to test if the CHAT-23-A can accurately distinguish ASD from other developmental delays or mental disorders, rather than just comparing ASD to typically developing children.
  • Economic Analysis: While the authors claim the system is cost-effective, a formal statistical comparison of the "cost-per-case-identified" between offline and online methods is still needed.

Summary updated

Abstract

The online screening system based on telemedicine technology is suitable for large-scale population ASD screening in childcare institutions.
BibTeX
@inproceedings{qiu2022application,
  author = {Ting Qiu and Heng Zhang and Conghua Zhou and Qilong Tang and Lizhen Wang and Xiaoyan Ke},
  title = {{Application of Telemedicine for Preliminary Screening of Autism Spectrum Disorder}},
  booktitle = {Frontiers in Pediatrics},
  year = {2022},
  doi = {10.3389/fped.2021.745597},
}
Figure