This paper reflects on the authors' experiences with three healthcare data visualization projects that employed a progressive data science approach. The projects focused on visualizing surgical outcomes, tracking patient bed transfers, and integrating patient-generated data into healthcare settings.
The authors highlight the key challenges faced at different stages of the progressive data science workflow:
Data Selection: Inconsistent data collection practices across healthcare centers, with some centers collecting limited data initially, posed challenges in the design and development process.
Pre-processing: Varying levels of data completeness, with some centers collecting additional response options beyond the initial dataset, required modifications to the visualization design and color schemes.
Transformation: The need to adapt database structures and recalculate metrics to accommodate new data formats introduced delays and required re-testing of previously developed components.
Data Mining: Displaying partial results during the progressive process led to misaligned expectations, as stakeholders anticipated seeing final calculations rather than intermediate outputs.
Interpretation and Evaluation: Revising calculation formulas and adapting visualizations to comply with healthcare system requirements introduced further challenges near the end of the projects.
The authors emphasize the need for careful consideration when using a progressive data science approach in healthcare visualization design. Developing standardized methods and tools to streamline data collection, pre-processing, and interpretation could help address these challenges and unlock the full potential of progressive data science in healthcare settings.
Sang ngôn ngữ khác
từ nội dung nguồn
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Faisal Zaki ... lúc arxiv.org 09-18-2024
https://arxiv.org/pdf/2409.10537.pdfYêu cầu sâu hơn