CITATION.cff

cff-version: 1.2.0
message: "If you use DeepCatch in your research, please cite it as follows."
title: "DeepCatch: Performance-Weighted Multi-Modal Fusion for Ultra-Early Cancer Detection from cfDNA"
authors:
  - given-names: "Royce"
    family-names: "[Last Name]"
    affiliation: "DeepCatch Consortium"
    orcid: ""
  - name: "DeepCatch Contributors"
year: 2026
doi: "to be assigned (preprint)"
url: "https://github.com/deepcatch/deepcatch"
repository-code: "https://github.com/deepcatch/deepcatch"
license: MIT
version: "1.0.0-preprint"
date-released: 2026-04-28
keywords:
  - "liquid biopsy"
  - "cfDNA"
  - "cancer screening"
  - "multi-modal fusion"
  - "longitudinal analysis"
  - "variant calling"
  - "early cancer detection"
  - "circulating tumor DNA"
  - "graph neural network"
  - "meta-learning"
abstract: >
  DeepCatch is a computational framework for pan-cancer detection at variant
  allele fractions as low as 0.001% — two orders of magnitude below current
  clinical detection limits. It integrates Bayesian contrastive variant calling,
  heterogeneous graph neural network (GNN) multi-modal fusion, cumulative
  evidence tracking (CET) from longitudinal blood draws, and meta-learning
  ensemble with 4-tier risk stratification.
identifiers:
  - type: doi
    value: "to be assigned (preprint)"
    description: Preprint DOI — to be assigned upon submission