stresspose-analysis

This repository contains the code for the project "Effect of acute psychosocial stress on body posture and movements".

This package contains various helper functions to work with the dataset (including tpcp Dataset representations) and to process the recorded data. Additionally, it contains analysis experiments performed with the dataset to reproduce all results, figures, etc.

Repository Structure

The repository is structured as follows:

bash ├── stresspose_analysis/ # `stresspose-analysis` Python package with helper functions └── experiments/ # Folder with conducted analysis experiments; each experiment has its own subfolder ├── pilotstudy/ # Data Analysis for the Pilot Study │ ├── feature_export/ # Exported motion features │ ├── notebooks/ # Notebooks for data processing, data analysis, classification experiments, plotting, etc. in subfolders │ ├── output/ # Output of classification experiments (pickled models, input features, etc.) │ ├── params/ # Hyperparameter settings for classification experiments │ ├── results/ # Analysis results │ └── config.json/ # Config file containing location of dataset (ignored because path depend on local configurations) └── mainstudy/ # Data Analysis for the Main Study ├── feature_export/ # Exported motion features ├── notebooks/ # Notebooks for data processing, data analysis, classification experiments, plotting, etc. in subfolders ├── output/ # Output of classification experiments (pickled models, input features, etc.) ├── params/ # Hyperparameter settings for classification experiments ├── results/ # Analysis results ├── scripts/ # Scripts for feature extraction and classification experiments └── config.json/ # Config file containing location of dataset (ignored because path depend on local configurations)

Installation

Dataset

In order to run the code, first download the desired dataset(s) from OSF: * Pilot Study * Main Study

Then, create a file named config.json in the respective experiment subfolders (/experiments/pilotstudy or /experiments/mainstudy) folder with the following content: json { "name-of-deployment <e.g., local>": { "base_path": "<path-to-dataset>" } }

This config file is parsed by all notebooks to extract the path to the dataset.
NOTE: This file is ignored by git because the path to the dataset depends on the local configuration!

Code

If you want to use this package to reproduce the analysis results the best way is to clone the repository and install the package via poetry. For that, open a terminal and run the following commands:

bash git clone git@github.com:empkins/stresspose-analysis.git cd stresspose-analysis poetry install # alternative: pip install .

This creates a new python venv in the stresspose-analysis/.venv folder. Next, register a new IPython kernel for the venv: bash cd stresspose-analysis poetry run poe register_ipykernel

Finally, go to the experiments folder and run the Jupyter Notebooks to reproduce all data processing steps (see below).

Experiments

Machine learning-based detection of acute psychosocial stress from body posture and movements

This analysis was performed for the paper "Machine learning-based detection of acute psychosocial stress from body posture and movements", currently under review at Scientific Reports.

Notebook Processing Order

To run the data processing and analysis pipeline, we recommend executing the notebooks in the following order: 1. Pilot Study: 1. Data Processing and Feature Extraction: 1. Motion Features: Run data_processing/Motion_Data_Preprocessing.ipynb -> data_processing/Motion_Data_Feature_Extraction.ipynb 2. Self-reports: Run data_processing/Questionnaire_Processing.ipynb 3. Saliva Data: Run data_processing/Saliva_Processing.ipynb 2. Classification Experiments: Run classification/Classification_General.ipynb and classification/Classification_Talk.ipynb 3. Data Analysis: 1. Saliva Data: Run stress_response/Saliva_Analysis.ipynb 2. Self-reports: Run stress_response/State_Questionnaire_Analysis.ipynb 3. Motion Features (statistical): Run motion_features/Motion_Feature_Overview.ipynb and motion_features/Motion_Feature_Analysis.ipynb 4. Classification Results: Run classification/Analysis_Classification_General.ipynb and classification/Analysis_Classification_Talk.ipynb 4. General Plots (plots used in papers, etc.): Run plotting/Teaserfigure_Plots.ipynb, plotting/Sensor_Grid_Plots.ipynb, and plotting/Filter_Pipeline_Plots.ipynb 2. Main Study: 1. Data Processing and Feature Extraction: 1. Motion Features: Run movement_data/feature_extraction/Feature_Extraction.ipynb and movement_data/feature_extraction/Feature_Extraction_per_Phase.ipynb (alternative: run the scripts scripts/feature_extraction.py and scripts/feature_extraction_per_phase.py) 2. Self-reports: Run questionnaires/Questionnaire_Processing.ipynb 3. Saliva Data: Run biomarker/Saliva_Processing.ipynb 2. Classification Experiments: Run the scripts (not notebooks!) scripts/classification.py and scripts/classification_rf.py. Due to the long runtime of the classification experiments, only a subset of classification pipeline permutations are executed in one script. Thus, the type of feature selection and scaler can be passed as environment variables. See the scripts for more information. 3. Data Analysis: 1. Saliva Data: Run analysis/stress_response/Saliva_Analysis.ipynb 2. Self-reports: Run analysis/stress_response/State_Questionnaire_Analysis.ipynb 3. Motion Features (statistical): Run analysis/movement_data/Movement_Feature_Overview.ipynb, analysis/movement_data/Movement_Feature_Analysis.ipynb, and analysis/movement_data/Movement_Feature_Analysis_per_Phase.ipynb 4. Classification Results: Run analysis/classification/Analysis_Classification_General.ipynb and analysis/classification/Analysis_Classification_Per_Phase.ipynb