SciDatareportR: A Comprehensive R Package for Streamlined Clinical Data Science

Introduction: : Clinical data science seeks insight into understanding disease mechanisms or developing successful therapeutics by integrating clinical, biomarker, and other relevant data. Achieving this goal requires rigorous data management—from validating disparate sources (e.g., electronic health records, biomarker panels) to systematically exploring variables that may predict clinical outcomes using different statistical methods. However, these steps often involve labor-intensive processes prone to errors and inconsistent documentation that may also lead to poor reproducibility of analysis.
In response, the Biomarker Core of the Johns Hopkins Center for the Advancement of HIV Neurotherapeutics(JH-CAHN) developed SciDataReportR, an open-source R package designed to streamline data cleaning, validation, and exploratory analysis in clinical settings. Originating from work on neurocognitive impairment in people living with HIV, SciDataReportR facilitates integrating clinical and biomarker data to uncover associations with outcomes such as cognitive decline. SciDataReportR delivers structured, modular pipelines and generates comprehensive reports, creating a reproducible foundation for downstream modeling and hypothesis testing. Its flexible architecture and customizable templates extend beyond HIV research—supporting a broad range of precision-medicine applications by enabling transparent, efficient integration of diverse clinical and biomarker data. SciDataReportR’s generalizable framework supports broader precision medicine efforts, enabling rapid deployment and reproducible data analysis across numerous clinical domains.

Materials and

Methods: : SciDataReportR (R ≥ 4.0) was was built in R (≥4.0) and leverages the tidyverse ecosystem to manage heterogeneous data sources, including patient demographics, laboratory results, and biomarker assays (e.g., metabolomics). The fully open-source code and detailed documentation are available on GitHub at https://github.com/RDastgh1/SciDataReportR.
Functions and suggested data analysis pipelines include a reproducible project scaffold for data organization within projects. All steps are rendered into version-controlled Quarto or R Markdown reports, providing reproducible documentation for downstream modeling and hypothesis testing. Custom wrapper functions are designed for streamlined and reproducible analysis with both supervised and unsupervised methods, including; exploratory data analysis, group comparisons, correlation matrices, and dimension reduction
To display the functions and capabilities of SciDataReportR we implemented our Data Analysis Pipeline (Fig1) on datasets with clinical, demographic, and biomarker data for neurocognitive impairment.

Results, Conclusions, and Discussions:: Applied to our neurocognitive-impairment dataset, SciDataReportR collapsed weeks of manual wrangling into hours. The project scaffold ensured proper data handling and its codebook enforcement eliminated inconsistent labels across visits, and data-quality checks immediately highlighted missing entries in key assays and a handful of extreme outliers, which we reviewed interactively.
The exploratory data analysis produced summary tables and figures of continuous and categorical data distributions and tables of group comparisons created publication-ready tables with our desired formatting. Correlation heatmaps revealed associations between continuous traits and biomarkers, and ANOVA heatmapes revealed associations between continuous biomarkers and categorical outcomes. Dimension reduction for the biomarkers was easily achieved using principal components analysis, with build-in visualizations for scree plots aiding in the selection of components and lollipop plots guiding interpretation of component loading. Built in functions for interactive and 3-dimensional plots helped visualize potential group differences.

These results underscore SciDataReportR’s capacity to facilitate precision medicine studies. By handling messy, real-world clinical datasets and automating key preprocessing steps, the package ensures reproducible workflows where study teams can confidently track data transformations. While initially developed for neurocognitive biomarker research in PLWH, SciDataReportR’s flexible design accommodates a broad range of clinical problems where integrated data pipelines are essential.

Overall, SciDataReportR addresses a crucial bottleneck in clinical data science and precision medicine: the lack of standardized, transparent workflows for merging clinical and biomarker data. The software’s modular approach unifies data cleaning, validation, and EDA within a single reproducible framework, potentially accelerating the identification of individualized risk factors and therapeutic targets. Future enhancements will expand its capabilities for handling specialized omics pipelines and advanced imputation strategies. We invite the BMES community to adopt and adapt SciDataReportR for diverse clinical applications, ultimately advancing the shared goal of delivering tailored healthcare interventions based on robust, data-driven insights.

Acknowledgements and/or References (Optional)::