| Autor: |
V. N. Kipen, M. V. Bahdanava, A. A. Burakova, O. I. Dobysh, O. V. Zotova, T. S. Koroleva, A. G. Bulgak, A. V. Baida, S. A. Bruskin, V. A. Lemesh |
| Rok vydání: |
2021 |
| Zdroj: |
Молекулярная и прикладная генетика. 30:39-48 |
| ISSN: |
1999-9127 |
| DOI: |
10.47612/1999-9127-2021-30-39-48 |
| Popis: |
Based on the data presented in the public domain on the GEO-NCBI Platform for 21 projects to determine the genomewide DNA methylation profile using the Infinium Human Methylation 450K BeadChip (Illumina©), we calculated the prediction error for chronological age values in percentage terms and MAD using three different algorithms: linear regression, generalized regression neural networks and multilayer feedforward neural networks. For each algorithm, we have provided guidelines to select the samples for the study. We have also proved that the minimum and sufficient sample size, an increase in which does not lead to a significant decrease in the MAD value, is at least 200 people. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
