Data Quality in Medical Real-World Data - An Oncological Use Case

Area of Science:

• Medical Informatics
• Oncology
• Data Science

Background:

• Real-world data (RWD) is crucial for data-driven medical research.
• Data quality deficiencies significantly limit the utility of RWD.
• Common issues include missing data, class imbalances, and timeliness problems.

Purpose of the Study:

• To identify and discuss common data quality deficiencies in real-world medical datasets.
• To analyze these deficiencies within an oncological use case.
• To improve the applicability of RWD in medical research.

Main Methods:

• Compiled a multi-departmental real-world dataset of 13861 cancer cases from University Hospital Cologne.
• Examined data quality throughout the data integration process.
• Focused on identifying missing data, class imbalances, and timeliness issues.

Main Results:

• Identified significant data quality deficiencies in the real-world cancer dataset.
• Observed issues such as missing patient information and imbalanced representation of cancer classes.
• Timeliness of data recording also presented challenges.

Conclusions:

• Data quality deficiencies are prevalent in real-world medical datasets, particularly in oncology.
• Addressing these issues is essential for reliable data-driven medical research.
• Improving data quality enhances the value of RWD for clinical insights and healthcare advancements.