Continuous Learning AI Model for Monitoring Autism Progress and Long-Term Developmental Outcomes: Sustainable Framework for Future-Oriented Autism Support
Posted: Aug 06, 2025
Abstract
Long-term monitoring of autism spectrum disorder progression requires adaptive systems capable of capturing complex developmental trajectories and evolving
support needs. This research presents a continuous learning artificial intelligence
framework designed for sustainable monitoring of autism progress across the lifespan, integrating multimodal data streams to predict long-term outcomes and dynamically adjust support recommendations. Our system employs online learning
algorithms that continuously update based on new observations, transfer learning techniques that preserve knowledge while adapting to individual progression
patterns, and uncertainty quantification methods that provide reliable confidence
estimates for clinical decision-making. The framework was developed and validated using longitudinal data from 4,280 individuals with autism spanning 2-22
years of age, encompassing 1.2 million data points across behavioral assessments,
intervention records, educational outcomes, and real-world functioning measures.
The continuous learning model achieved 93.7% accuracy in predicting 12-month developmental trajectories and 91.2% accuracy in forecasting long-term outcomes up
to 5 years, significantly outperforming static models while maintaining computational efficiency. Implementation across 35 clinical sites demonstrated sustainable
performance improvement, with prediction accuracy increasing by 18.4% over 24
months through continuous learning from new cases. The system successfully identified critical transition points in development, predicted regression periods with
87.9% accuracy, and provided early warnings for support need changes 8.3 months
before clinical recognition. This research establishes a sustainable, future-oriented
approach to autism progress monitoring that evolves with accumulating knowledge
and changing individual needs, representing a paradigm shift from static assessment
to dynamic, lifelong support systems.
Downloads: 153
Abstract Views: 318
Rank: 222966