An intelligent fuzzy-neural framework for autism sensory assessment using hierarchical linguistic modeling and risk-based temporal decision-making

Autistic diagnosis and sensory tests present subjectivity, temporal behaviors, and vagueness. In the autistic sensory classification model, Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks are considered for temporal sensory response patterns learning. To alleviate these challenges, we introduced an original hybrid framework that combines double hierarchy hesitant linguistic term sets (DHHLTS), temporal Three-Way Decision-Making (TWD). There are essentially three advantages of our research work in that, first, double fuzzy hierarchy mapping for dealing with precise expert evaluations. Secondly, temporal-aware RNN architecture is motivated by Hamacher t-norm/t-conorm aggregations. Lastly, explainable probabilistic TWD for risk categorization. This work combines fuzzy logic and decision theory to furnish an executable tool for caring for autistic people.