A serverless application for human stress detection in and through sleep, hosted with Google Cloud Platform. Note: this is NOT a commercial project! All rights, including data and technology, belong to the SaYoPillow organization. The project does not violate the terms and agreements provided by the author's license.
The dataset page on Kaggle. The goal of the research is to solve a supervised classification problem for predicting stress level using numerical features (multilabel classification problem). Featues used for classification: snoring range of the user (sr), respiration rate (rr), body temperature (t), limb movement rate (lm), blood oxygen levels (bo), eye movement (rem), number of hours of sleep (sh) and heart rate (hr). The label represents a certain stress level: 0 - Low or normal, 1 – Medium low, 2 - Medium, 3 - Medium high, 4 - High.
GCP functions can be invoked locally using the Google Cloud SDK for Python, specifically the functions-framework package, which is already included to the dependencies list. To run Classifier HTTP Trigger use following command:
functions-framework --target calculate_stress_level --debug