From Sparse Visits to Continuous Molecular Trajectories with Flow Matching
Longitudinal transcriptomic studies often capture only a few molecular snapshots of each patient over time. This makes it challenging to understand how patients are progressing, compare them when progression rates differ, and capture molecular changes between observed visits. We introduce CohortFM, a flow-matching framework that reconstructs continuous patient-specific molecular trajectories from…
Research introduces CohortFM, a framework to reconstruct continuous molecular trajectories from sparse longitudinal observations. This tool organizes patient-specific trajectories through a mixture model, allowing researchers to analyze progression patterns without calendar time dependence. When evaluated across COVID-19, sepsis, influenza, and tuberculosis, CohortFM consistently generates meaningful trajectories despite varying sampling patterns.
These trajectories reveal biological differences between patients that cannot be discerned from observed samples alone. Compared to existing trajectory baselines, CohortFM yields more biologically distinct trajectories and a more comprehensive representation of the patient population. Its finer progression scale uncovers biological programs and temporal dynamics missed by visit-level analysis.
The framework also demonstrates strong interpolation capabilities and forecasting accuracy. By employing flow matching as its core learning mechanism, CohortFM improves continuous trajectory learning. The study concludes that CohortFM is a practical tool for converting sparse molecular snapshots into continuous, biologically interpretable trajectories, enabling disease progression studies and future molecular state predictions even with few patient samples.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
