Abstract
BioPsyKit is a comprehensive Python package designed to streamline the analysis of multi-modal biopsychological data. It provides a unified framework for processing physiological signals, movement data, biochemical markers, and psychological questionnaires. By integrating gold-standard tools into a modular architecture, BioPsyKit addresses the fragmentation of research software, enhancing reproducibility and accessibility in the field.
Introduction
Biopsychological research is inherently complex, requiring the integration of data across vastly different scales—from high-frequency physiological signals (ECG, EEG) to biochemical markers (cortisol) and subjective self-reports. Historically, the lack of a single tool to handle these diverse data types has led to "library hopping" and the creation of non-reproducible custom scripts. BioPsyKit fills this critical gap by providing a "one umbrella" philosophy for data analysis.
Methods and Architecture
BioPsyKit is organized into several core modules that handle specific data types and experimental requirements. It builds upon existing libraries like NeuroKit2 for low-level signal processing while adding specialized layers for outlier removal and plotting.
Key modules include:
signals: Handles (electro)physiological data.sleep: Implements algorithms to derive sleep endpoints from accelerometer data.saliva: Calculates biomarker characteristics like Area Under the Curve (AUC).questionnaires: Automates the scoring of standard psychological scales.protocols: Represents experimental designs (e.g., TSST, MIST) as data structures to align multi-modal data.
Key Insights and Results
The primary innovation of BioPsyKit is the integration of data streams. By treating experimental protocols as first-class objects, the software helps researchers maintain context throughout the analysis pipeline. This reduces boilerplate "glue code" and ensures that different labs calculate metrics in a standardized way.
Furthermore, the package includes high-level plotting functions that simplify the creation of publication-quality figures with automatic statistical annotations. This is particularly useful for visualizing biomarker responses during stress protocols like the Trier Social Stress Test (TSST).
Conclusion
BioPsyKit successfully bridges the gap between raw data and scientific insight. While it currently relies on external dependencies and focuses on post-hoc analysis, it provides a robust foundation for reproducible biopsychological research. Future work may involve deeper machine learning integration and real-time analysis capabilities.
Summary updated