Application of Artificial Intelligence in Screening, Diagnosis, and Intervention Rehabilitation Services for Children with Autism Spectrum Disorder
Childhood Autism, also known as autism spectrum disorder (ASD), is a neurodevelopmental disorder that affects children’s social interaction, communication, and behavior, typically manifesting before the age of three. Children with this condition may encounter challenges in establishing interpersonal relationships, expressive language, and daily adaptation, while often exhibiting intense interests in specific subjects and heightened sensitivity to changes.
The diagnosis of autism relies on comprehensive evaluations by professional teams. Although there is currently no cure, early intervention measures—such as behavioral therapy and training in language and social skills—can significantly improve patients’ quality of life. Societal understanding and support for children with autism are crucial, and ongoing scientific research aims to provide greater assistance to these children and their families.
Recently,At the Top 100 Summit of the VBEF Future Medical Ecosystem Expo hosted by VCBeat, Zhang Huai, Deputy Director of the Nanjing Qingzhan Artificial Intelligence Research Institute, delivered a speech titled “Applications of Artificial Intelligence in Screening, Diagnosis, Intervention, and Rehabilitation Services for Childhood Autism.”, the following is a transcript of the speech.
From the perspective of Dean Zhang Huai, the application of artificial intelligence in screening, diagnosis, and intervention/rehabilitation services for children with autism is an innovative endeavor that integrates medicine and engineering. It brings together multidisciplinary expertise—including clinicians, rehabilitation specialists, AI algorithm developers, and hardware and software designers—offering new hope to children with autism. The Qingzhan Institute of Artificial Intelligence aims to apply AI technologies in the healthcare sector to enhance screening, diagnosis, and rehabilitation services for children with autism.
According to official statistics, the prevalence of autism in China is 7 per 1,000, but the actual figure may be higher. Due to insufficient awareness among parents and society, the diagnosis rate remains relatively low. Dean Zhang believes that the manifestations of autism are diverse; some are easily quantifiable, while others require qualitative analysis. The current challenge lies in translating these characteristics into programmable and standardized criteria for detection and diagnosis.
Regarding screening, President Zhang stated, “I have always emphasized ‘healthcare plus artificial intelligence,’ rather than ‘artificial intelligence plus healthcare.’” He advocates for healthcare to take the lead, with AI technology providing subsequent support. This means that clinicians should first define the requirements and standards, after which engineers and technical personnel develop corresponding technological solutions under the guidance of these specifications. Within this framework, Qingzhan Artificial Intelligence Research Institute strictly adheres to these standards, aiming to achieve early detection and early diagnosis of childhood autism.
In the diagnostic process for autism, rating scales serve as one of the diagnostic criteria; their ease of quantification and standardization makes the diagnostic procedure relatively straightforward. However, not all diagnostic methods are this simple, particularly in the diagnosis of childhood autism, where behavioral characteristics such as a child’s neglect of or lack of response to surrounding stimuli are difficult to quantify. Determining the specific extent to which a patient fails to look or listen, and discerning whether such behavior stems from a lack of interest or other factors such as visual impairments, poses a challenge not only for clinicians but also for the application of artificial intelligence.
Currently, existing technologies such as human behavioral posture sensing and data analysis have demonstrated potential in the screening and diagnosis of autism. Although clinicians can make rapid judgments based on experience and intuition, quantitative data are increasingly needed to support decision-making during disease progression or treatment. Director Zhang and his team aim to leverage these technologies to conduct accurate assessments across diverse settings—whether in hospitals, rehabilitation centers, or homes—with the goal of achieving optimal therapeutic outcomes.
Currently, the Qingzhan Artificial Intelligence Research Institute is focused on developing hardware and software solutions for autism assessment. These devices and methodologies have initiated testing in controlled environments, such as laboratories and rehabilitation centers. While international companies have achieved significant milestones in this field and obtained FDA clearance, there are currently no Class III medical device approvals in China specifically for autism and depression. In terms of intervention and rehabilitation, the institute’s existing achievements represent a seamless integration of traditional medical approaches with modern machine learning technologies.
President Zhang introduced that the Qingzhan Institute of Artificial Intelligence has partnered with Fangyuan Children’s Rehabilitation Center and Beijing Xinfeng Future Technology Co., Ltd. to integrate clinical diagnosis with rehabilitation practice. One of the developed products enables children with autism to engage in interactive activities tailored to their individual needs, such as drawing scenes depicting conversations with teachers. Key features of the product include high customizability and real-time responsiveness, allowing it to deliver personalized experiences adapted to different developmental stages and scenarios. This interactivity and real-time feedback facilitate dynamic adjustments to children’s responses and treatment plans.
Furthermore, corresponding AI products, such as the Linglong and Fangyuan teaching software, have been implemented in frontline educational settings, enhancing existing textbooks and teaching aids. These tools primarily leverage AI technologies, including text-to-image and image-to-text generation, to support instruction. Throughout this process, we place significant emphasis on privacy protection and localized deployment to ensure the privacy and security of children.
Currently, the Qingzhan Institute of Artificial Intelligence has received substantial positive feedback, with stakeholders recognizing that this teaching methodology exhibits high flexibility and generalizability. It is capable of adapting to the diverse developmental stages of children aged 1 to 6 years, achieving favorable outcomes under the guidance of educators and medical diagnostics. Furthermore, the team is exploring interactions between the digital and physical worlds, such as leveraging AI solutions to enhance efficiency and alleviate parental stress. In medical applications, reliability and interpretability are paramount. Dean Zhang stated that they are considering integrating visual cues, auditory signals, and various physiological indicators to address issues related to interpretability and safety. Similarly, there is an aspiration to develop companion robots and provide caregiving services, thereby further advancing development in this field.
In his closing remarks, President Zhang Huai shared, “We have created the ‘Star Children’ world for our pediatric patients. These initiatives have been warmly received by the young patients, giving us a profound sense of accomplishment. We are eager to collaborate with more enterprises to better serve patients and translate our technologies into practical applications.”