Genome-wide prediction method based on deep learning by using genome-wide data and bioinformatics features
US12315600B2 · kind B2 · utility
Assignees
Inventors
Key dates
| Filing date | Feb 8, 2024 |
| Grant date | May 27, 2025 |
| Priority date | — |
| Expiry date | Feb 8, 2044 |
Classification
- Technology area (CPC Y)Emerging Cross-Sectional Technologies
- CPC primaryY02A90/10
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Genome-wide data is obtained, and data cleansing, data sparsity processing and bioinformatics feature extraction are performed on the obtained genome-wide data; model construction is performed based on the sparsity-processed genome-wide data and the bioinformatics features to obtain a preliminary hybrid model; model training, regularization, and interpretability enhancement are performed on the preliminary hybrid model to obtain a trained model weight and an interpretability analysis corresponding to the trained model weight; learning and uncertainty estimation are performed based on the trained model weight and to-be-predicted genome-wide data on the hybrid model to obtain an integrated prediction result and uncertainties corresponding to the integrated prediction result; and personalized medical advice and decision assistance are performed based on the integrated prediction result, the interpretability analysis, and the uncertainties corresponding to the integrated prediction result.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.