CohortFinder

CohortFinder improves machine learning model generalizability in digital pathology and medical imaging by intelligently partitioning data cohorts to avoid bias from batch effects.

Key Features:

• Data-driven cohort partitioning: CohortFinder mitigates batch effects via data-driven partitioning of digital pathology and imaging cohorts.
• Batch effect mitigation: CohortFinder addresses systematic technical differences in data collection unrelated to biological variations that negatively impact machine learning model generalizability.
• Performance improvement: CohortFinder improves machine learning model performance in downstream digital pathology and medical image processing tasks.

Scientific Applications:

• Digital pathology: CohortFinder improves machine learning model generalizability in digital pathology tasks.
• Medical imaging: CohortFinder improves machine learning model generalizability in medical image processing tasks.

Methodology:

CohortFinder uses data-driven cohort partitioning to mitigate batch effects.