MEDSCAN Smartphone Image Analysis for Schistosomiasis Diagnostics

Observations from a multi-site laboratory evaluation of a smartphone application for interpreting point-of-care circulating cathodic antigen (POC-CCA) tests for schistosomiasis. The dataset includes cleaned data from smartphone-based image analysis and benchtop lateral flow reader measurements across three research sites. It supports the reproduction of statistical analyses, including ROC curves and agreement metrics, from the associated publication.

Use Cases

• Compare automated test line intensity readings from smartphone images against benchtop lateral flow reader measurements to validate the MEDSCAN application.
• Analyze agreement metrics, such as Cohen's Kappa, between different users (e.g., 'KEMRI-1', 'LUMC-2') and device types for POC-CCA test interpretation.
• Reproduce ROC analyses to evaluate the diagnostic performance of the smartphone-based image analysis method using the provided Python scripts and cleaned datasets.
• Investigate intraclass correlation coefficients for test results across the three laboratory sites (KEMRI, LUMC, VU) included in the study.

Strengths

• Data originates from a controlled multi-site laboratory evaluation involving 3 distinct research institutions.
• Includes complete reproduction materials, such as Python code and statistical results, enabling verification of the published findings.
• Data is provided in both Parquet and CSV formats, offering flexibility for analysis.

Limitations

• The dataset is small at 6.8 MB, limiting its utility for training large-scale machine learning models.
• The sample size in terms of total observations or rows is not specified, making it difficult to assess statistical power for new analyses.
• Data is specific to a single diagnostic method (POC-CCA) for one disease (schistosomiasis), limiting generalizability.

Provenance

Source

Thomas Scherr via figshare.

Collection Method

Collected during a multi-site laboratory evaluation of the MEDSCAN smartphone application, involving 9 users and multiple device types.

Freshness

Last updated in March 2026, indicating recent maintenance.

Geography

Data collected from three sites: KEMRI, LUMC, and VU.