arXiv:2607.07725v1 Announce Type: new Abstract: Genomic prediction models often fail to transfer across institutions because sequencing panels differ across sites, creating structural feature missingness at deployment. Existing approaches to this challenge typically restrict analysis to genes shared across cohorts, exclude patients with incomplete profiles, or rely on test-time imputation, all of which can reduce robustness and limit the use of multi-center data. We propose Survival prediction Handling Incomplete Features using Transformer (SHIFT), a missingness-aware survival model that direc
The proliferation of genomic data across diverse institutions creates an urgent need for models that can handle heterogeneity and incompleteness without sacrificing robustness.
Improving survival prediction from genomic data has significant implications for personalized medicine, drug discovery, and public health, especially when combining multi-center datasets.
This advancement changes how researchers can leverage incomplete genomic datasets from various sources, making multi-institutional collaborations more effective and predictions more accurate without restrictive data harmonisation.
- · AI in healthcare
- · Genomic data companies
- · Biopharmaceutical R&D
- · Oncology patients
- · Traditional statistical methods for incomplete data
- · Institutions with proprietary, non-interoperable data standards
More accurate and reliable survival predictions in clinical settings using diverse genomic data.
Accelerated development of personalized cancer therapies and prognostic tools due to better data utilization.
Potential for a global network of interoperable genomic prediction models, leading to new insights into disease mechanisms and population health.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG