Interpretable Neuro-Fuzzy Models for Stress Prediction.

Machine learning algorithms are useful in assisting medical judgments, but the resulting models are frequently challenging for doctors to comprehend. Conversely, IF-THEN rules articulated in natural language are successful at describing interpretable models that are returned by Fuzzy Inference Systems. However, the construction of the fuzzy rule basis in conventional fuzzy systems requires human expertise. Neuro-fuzzy systems can infer fuzzy rule-based models from data, saving time by doing away with the need to manually build the rules. Using neuro-fuzzy systems to develop prediction models from data in the form of fuzzy rules that are suitable to enhance decision-making for stress assessment is what we propose in this paper. Our results highlight how well neuro-fuzzy models perform in providing precise predictions while maintaining interpretability.