AI In Medicine: University of Southern California Develops AI Platform That Helps In The Early Diagnosis Of Autism Via A Five-Minute Ipad Coloring Game!
Nikhil Prasad Fact checked by:Thailand Medical News Team Dec 21, 2023 11 months, 1 day, 10 hours, 1 minute ago
AI In Medicine: In a monumental leap forward for the field of healthcare, researchers at the University of Southern California (USC) have harnessed the power of artificial intelligence (AI) to enhance the early diagnosis of autism. Led by Professor Dr Lisa Aziz-Zadeh, the team's groundbreaking study introduces an innovative AI platform that utilizes data from a five-minute iPad coloring game.
A Five Minute Ipad Based Colouring Game That Uses AI To Diagnose Autism
This technology covered in this
AI In Medicine news report, not only shows remarkable accuracy in distinguishing between typically developing children and those with autism spectrum disorder (ASD) but also opens new avenues for early detection and intervention.
The Intersection of AI and Autism
The intricate nature of autism spectrum disorder poses unique challenges in early detection, often overlapping with other developmental disorders such as developmental coordination disorder (DCD). Recognizing the critical need for early identification, the USC study pioneers the use of machine learning analytics, a subset of AI, to process data collected from a specially designed iPad coloring game.
The Study's Methodology
Engaging 54 children aged 8 to 17, including 18 with ASD, 16 with DCD, and 20 typically developing children, the study utilized iPads to collect touchscreen kinematic data. This data encompassed factors such as pressure applied and the smoothness of movements during a five-minute coloring game. Machine learning algorithms then processed this nuanced data to distinguish between the different groups. Impressively, the AI platform achieved an accuracy of 76% in distinguishing between typically developing children and those with ASD, 78% for distinguishing between typical development and DCD, and 71% for distinguishing between ASD and DCD.
The Significance of Early Detection
Early identification of developmental disorders is paramount for implementing tailored therapeutic interventions, resulting in improved long-term developmental outcomes. Dr Christiana Dodd Butera, the study's first author, underscores the goal of providing appropriate therapy at the most impactful time in development. Identifying motor signatures associated with autism, particularly before the emergence of social symptoms, represents a significant stride in early detection without the risk of bias introduced by human assessors.
Understanding Motor Signature Differences
The study delves into the motor signature differences between ASD and DCD, shedding light on the distinct clinical features of these neurodevelopmental disorders. Motor differences have long been observed in individuals with ASD, even in the neonatal period, suggesting prenatal neurodevelopmental origins. In contrast, DCD is characterized by impairments in fine and gross motor skills, underscoring the need for precise diagnostic tools to differentiate between these disorders.
Kinematics and Neural Mechanisms
rong>
The researchers employed machine learning computational analysis to identify kinematic markers that strongly differentiate between ASD and DCD. The study also explored neural correlates using functional magnetic resonance imaging (fMRI) during action production tasks. The cerebellum emerged as a key region associated with motor differences in both ASD and DCD groups, providing valuable insights into the neurobiological basis of these disorders.
Cerebellar Involvement
The cerebellum, implicated in various motor and cognitive functions, exhibited functional and structural differences in both ASD and DCD groups. Alterations in cerebellar gray matter structure, disrupted white matter tracts, abnormal functional connectivity, and altered activity during motor tasks were observed in individuals with ASD and DCD. The study underscores the relevance of cerebellar regions in understanding the interplay between sensorimotor function and cognitive processing.
Motor Games and Machine Learning
Traditional assessments of motor skills can be limited in capturing subtle coordination and timing differences. The study explores the use of machine learning on data from a smart tablet motor game to distinguish between ASD and DCD. The kinematic markers derived from the digital game proved effective in differentiating between clinical groups, highlighting the potential of AI-driven technology in early identification and diagnosis.
Classifying ASD/DCD/TD by Game-Play
Coupled with machine learning, kinematics recorded from the smart tablet game were able to categorize ASD from TD at 76%, ASD from DCD at 71%, and DCD from TD at 78% accuracy. This is the first time serious game digital technology has been used to distinguish two similar motor developmental disorders-ASD from DCD. Standard behavioral motor measures and video coding analysis could not distinguish the two groups apart, making this finding especially promising. The method may contribute to clinical diagnosis and better inform the particular underlying motor disturbances in each group, though further research with larger sample sizes is essential. Refinements of this technique can be explored in future studies, potentially including social motor games.
Motor Markers that Distinguish Groups
The kinematic markers that most contribute to differentiating between groups include the control of deceleration and variability in the distance, or area covered, of the motor gestures. Autistics exhibited more variability in the size of the gesture area used on the smart tablet than individuals with DCD for each motor gesture. Future work will need to investigate this to better understand individual action patterns and their distribution in autistics.
Neural Correlates and Cerebellar Regions
The study found that cerebellar regions, specifically crus I/II, were associated with the kinematic features that best differentiated between specific pairwise groups. Activity in the left crus II correlated with Gesture Directness Variance across participants and with Gesture Area Variance in the ASD group during motor tasks. These data indicate that during motor tasks, the right and left crus II are particularly hypoactive in DCD, while activity patterns in crus I may be more nuanced between groups. Differential activity in these cerebellar regions may lead to behavioral motor differences between groups, allowing the use of kinematic patterns to distinguish between ASD, DCD, and TD groups.
The Integration of AI and Smart Technology
The convergence of smart technology, exemplified by iPads, with AI signifies a paradigm shift in autism research. The study highlights the potential of machine learning analytics to process nuanced kinematic data obtained from digital platforms. This approach not only streamlines the diagnostic process but also opens avenues for further research into the intersection of technology and healthcare, showcasing the transformative potential of AI in neurodevelopmental disorder research.
Challenges and Future Directions
While the study demonstrates promising results, the researchers acknowledge the need for replication in larger, more diverse groups of children. The current study focused on high-functioning children and adolescents, underscoring the necessity for further exploration in younger populations. The researchers emphasize the importance of identifying motor signatures as early as possible, indicating the potential for this technology to contribute to early intervention strategies that can significantly impact the lives of affected children.
Conclusion
In conclusion, the USC study represents a significant stride in leveraging AI and smart technology for the early diagnosis of autism and developmental coordination disorder. The integration of machine learning analytics with a simple motor coloring game on a smart tablet demonstrates the potential to revolutionize the diagnostic landscape for neurodevelopmental disorders. While the study's findings are promising, further research and replication with larger and more diverse samples are essential for establishing the robustness and generalizability of this innovative approach. As AI continues to play a transformative role in healthcare, the USC researchers' work paves the way for more accessible, accurate, and early detection of developmental disorders, ultimately contributing to improved outcomes for affected children.
The study findings were published in the peer reviewed Journal of Autism and Developmental Disorders.
https://link.springer.com/article/10.1007/s10803-023-06171-8
For the latest on AI-In Medicine, keep on logging to Thailand Medical News.
Read Also:
https://www.thailandmedical.news/articles/ai-in-medicine
Share
Tweet
Share
