Prediction of lncRNA–cervical cancer associations using a multi-view collaborative variational autoencoder with knowledge distillation transfer learning
- 1Xinjiang Medical University Cancer Hospital, Urumqi, Xinjiang, China
- 2Kecheng District People’s Hospital, Quzhou, Zhejiang, China
Abstract
IntroductionLong non-coding RNAs (lncRNAs) play critical regulatory roles in the pathogenesis of cervical cancer, and accurate identification of lncRNA–cervical cancer associations is important for understanding the molecular mechanisms of the disease and discovering potential therapeutic targets. However, the extreme scarcity of cervical cancer-specific lncRNA association data constitutes a major bottleneck limiting the performance of existing computational prediction methods.MethodsTo address this challenge, we propose a Multi-View Collaborative Variational Autoencoder based on Knowledge Distillation Transfer Learning (MVCVAE), which integrates multi-source heterogeneous data fusion and cross-disease knowledge transfer in a task-oriented manner. The multi-view collaborative variational autoencoder constructs dual-view feature representations of lncRNA and disease similarities and employs a symmetric encoder–decoder architecture to learn complex lncRNA–disease association patterns in the latent space. The knowledge distillation transfer learning framework adopts a teacher–student network design: a teacher model is pre-trained on a large-scale dataset containing 12,865 lncRNA–disease associations to learn general association patterns, and the resulting prior knowledge is effectively transferred to a student model targeting 92 cervical cancer associations through feature distillation and prediction distillation strategies.ResultsIn five-fold cross-validation, MVCVAE achieves an AUC of 0.9228 ± 0.0058.DiscussionThrough the synergistic effects of multi-view heterogeneous information fusion and cross-domain knowledge transfer, MVCVAE demonstrates strong performance in predicting lncRNA–cervical cancer associations, effectively addressing the limitations imposed by the scarcity of cervical cancer-specific lncRNA association data.
Related
- Prediction of Obstructive Sleep Apnea Using Variational Autoencoder-Derived Features From 12-Lead ECG
- Real-time Evaluation and Anticipation of Causal disTillation early prediction for cardiac surgery-associated acute kidney injury (REACT-01): study protocol for a single-center prospective randomized clinical trial
- Perioperative Intravenous Antibiotic Regimens and Surgical Site Infection After Colorectal Surgery: Clinical Associations and External Evaluation of Risk Prediction Models
- Cardiovascular disease prediction using variational recurrent autoencoders with uncertainty estimation
- From Nutritional Epidemiology to Clinical Prediction: Interpretable Machine Learning Reveals Subtle Associations Between Dietary Antioxidant Profiles and Non-Melanoma Skin Cancer in Older Adults.
BibTeX
@article{yuan2026prediction,
title = {{Prediction of lncRNA–cervical cancer associations using a multi-view collaborative variational autoencoder with knowledge distillation transfer learning}},
author = {Min Yuan and Tingchuan Xiong and Xiariwana Abasi and Runmei Guo and Yandan Xu and Guzhalinuer Abulizi},
journal = {Frontiers in Genetics},
year = {2026},
month = nov,
doi = {10.3389/fgene.2026.1937699},
url = {https://doi.org/10.3389/fgene.2026.1937699},
note = {CC BY},
}
RIS
TY - JOUR
TI - Prediction of lncRNA–cervical cancer associations using a multi-view collaborative variational autoencoder with knowledge distillation transfer learning
AU - Min Yuan
AU - Tingchuan Xiong
AU - Xiariwana Abasi
AU - Runmei Guo
AU - Yandan Xu
AU - Guzhalinuer Abulizi
PY - 2026
DA - 2026/11/04
JO - Frontiers in Genetics
DO - 10.3389/fgene.2026.1937699
UR - https://doi.org/10.3389/fgene.2026.1937699
AB - IntroductionLong non-coding RNAs (lncRNAs) play critical regulatory roles in the pathogenesis of cervical cancer, and accurate identification of lncRNA–cervical cancer associations is important for understanding the molecular mechanisms of the disease and discovering potential therapeutic targets. However, the extreme scarcity of cervical cancer-specific lncRNA association data constitutes a major bottleneck limiting the performance of existing computational prediction methods.MethodsTo address this challenge, we propose a Multi-View Collaborative Variational Autoencoder based on Knowledge Distillation Transfer Learning (MVCVAE), which integrates multi-source heterogeneous data fusion and cross-disease knowledge transfer in a task-oriented manner. The multi-view collaborative variational autoencoder constructs dual-view feature representations of lncRNA and disease similarities and employs a symmetric encoder–decoder architecture to learn complex lncRNA–disease association patterns in the latent space. The knowledge distillation transfer learning framework adopts a teacher–student network design: a teacher model is pre-trained on a large-scale dataset containing 12,865 lncRNA–disease associations to learn general association patterns, and the resulting prior knowledge is effectively transferred to a student model targeting 92 cervical cancer associations through feature distillation and prediction distillation strategies.ResultsIn five-fold cross-validation, MVCVAE achieves an AUC of 0.9228 ± 0.0058.DiscussionThrough the synergistic effects of multi-view heterogeneous information fusion and cross-domain knowledge transfer, MVCVAE demonstrates strong performance in predicting lncRNA–cervical cancer associations, effectively addressing the limitations imposed by the scarcity of cervical cancer-specific lncRNA association data.
SN - 1664-8021
ER -