Fuzzy Legal Evaluation in Telehealth via Structured Input and BERT-Based Reasoning

This study proposes a structured approach for handling ambiguous user input in telehealth consent scenarios. A mobile application collects task-based entries, which are then evaluated using a publicly available BERT-based ambiguity detection model trained to detect semantic uncertainty. The model is based on the BERT architecture, allowing contextual interpretation of short, user-generated text. The model’s output is converted into clarity scores and mapped to fuzzy membership values, allowing integration into rule-based reasoning. Initial results suggest that the system can represent partial understanding and assist legal decision-making when inputs are vague or incomplete.