SynchDP: A correlation based sequence alignment algorithm for synchronizing longitudinal clinical data.

Area of Science:

• Computational Biology
• Bioinformatics
• Data Science

Background:

• Clinical time-series data present challenges due to irregular intervals and varying lengths.
• Comparing patient disease progression is difficult with asynchronous and irregularly sampled data.
• This hinders accurate analysis of gene expression profiles and biomarker discovery.

Purpose of the Study:

• To develop a novel algorithm for aligning and synchronizing multiple clinical time-series.
• To enable de novo assembly of representative patterns from input sequences.
• To facilitate robust comparison of patient disease trajectories.

Main Methods:

• A correlation-based dynamic programming algorithm, SynchDP, was developed.
• The algorithm handles irregular and asynchronous time-series data.
• SynchDP was validated using synthetic data and COVID-19 patient severity levels.

Main Results:

• SynchDP demonstrated superior alignment quality compared to existing methods on synthetic data.
• The algorithm successfully synchronized COVID-19 patient data, identifying health deterioration and recovery patterns.
• Biomarkers related to disease severity progression were effectively captured using single-cell transcriptome data.

Conclusions:

• SynchDP provides a robust method for aligning complex clinical time-series.
• The algorithm aids in understanding disease progression and identifying relevant biomarkers.
• This approach has significant implications for personalized medicine and biomarker discovery.