DeepOmicsSurv: a deep learning-based model for survival prediction of oral cancer Deepali, Neelam Goel, Padmavati Khandnor Discover Oncology, 2025 OBJECTIVE: Oral cancer is an important health challenge worldwide and accurate survival time prediction of this disease can guide treatment decisions. This study aims to propose a deep learning-based model, DeepOmicsSurv, to predict survival in oral cancer patients using clinical and multi-omics data. METHODS: DeepOmicsSurv builds on the DeepSurv model, incorporating multi-head attention convolutional layers, dropout, pooling, and batch normalization to boost its strength and precision. Various dimensionality reduction techniques, including Principal Component Analysis (PCA), Kernel PCA, Non-Negative Matrix Factorization (NMF), Singular Value Decomposition (SVD), Partial Least Squares (PLS), Multidimensional Scaling (MDS), and Autoencoders, were employed to manage the high-dimensional omics data. The model's performance was evaluated against DeepSurv, DeepHit, Cox Proportional Hazards (CoxPH), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN). Additionally, SHapley Additive Explanations (SHAP) was used to analyze the impact of clinical features on survival predictions. RESULTS: DeepOmicsSurv achieved a C-index of 0.966, MSE of 0.0138, RMSE of 0.1174, MAE of 0.0795, and MedAE of 0.0515, outperforming other deep learning models. Among various dimensionality reduction techniques, autoencoder performed the best with DeepOmicsSurv. SHAP analysis showed that Age, AJCC N Stage, alcohol history and patient smoking history are prevalent clinical features for survival time. CONCLUSION: In conclusion, DeepOmicsSurv has the potential to predict survival time in oral cancer patients. This model achieved high accuracy with various data types including Clinical, DNAmethylation + clinical, mRNA + clinical, Copy number alteration + clinical, or multi-omics data. Additionally, SHAP analysis reveals clinical factors that influence survival time.
Optimized Prognostic Models for Oral Cancer Survival using Feature Selection Methods Deepali, Neelam Goel, Padmavati Procedia Computer Science, 2024 Accurately predicting survival time holds significant importance in various medical and scientific domains. This study proposes a novel survival prediction model for oral cancer using multi-omics data including clinical, DNA methylation, copy number alteration, and mRNA data. The proposed model incorporates four feature selection methods, namely LASSO, Elastic Net, Random Forest Regressor, and Recursive Feature Elimination (RFE), to improve the accuracy. The findings reveal that RFE outperforms the other methods, achieving the highest C-Index of 0.944. This is a significant improvement over previous models, which have a c-index of 0.694 using only clinical data and 0.916 using multi-omics data. In terms of precision, the Random Forest Regressor is the most precise with an MSE of 18.37, followed by RFE of 21.61. Notably, both RFE and Random Forest Regressor outperformed LASSO and Elastic Net in predictive accuracy. Additionally, the results highlight the consistency of RFE, as indicated by its low Median Absolute Error (MedAE) of 2.86. Random Forest Regressor also performed well in this regard, with a MedAE of 4.05. In contrast, Elastic Net exhibited a higher MedAE of 4.55, while LASSO showed more variability, with a MedAE of 8.58. These findings offer valuable insights into survival analysis and feature selection, helping researchers and practitioners select appropriate approaches for survival time prediction, and improving accuracy and reliability in critical applications.
Link state and forwarding nodes constraint based DBR in UWSN (LCDBR) Harpinder Kaur, Deepali Goyal 2016 5th International Conference on Reliability Infocom Technologies and Optimization Icrito 2016 Trends and Future Directions, 2016 Underwater Wireless Sensor Network (UWSN) is emerging lots of attention due to its various types of characteristics and applications. Distinctive underwater characteristics leads to many challenges such as-high propagation delay, data losses, nodes mobility, localization, nodes deployment while designing any reliable and efficient routing protocol. In this paper, an extension of CDBR (Forwarding nodes Constraint based Depth Based Routing) protocol is proposed named LCDBR (Link state and Forwarding nodes Constraint based Depth Based Routing). In this protocol, packets are sent through the optimal path towards sink because no. of sinks is placed on the surface of water. Also, in multihop transmission forwarding nodes are constrained to only one for energy conservation. Simulations results are conducted through MATLAB which shows that the protocol outperforms CDBR protocol in terms of energy efficiency, packet throughput and end to end delay.
Improvement and analys security of WSN from passive attack Gagandeep Kaur, Deepali, Rekha Kalra 2016 5th International Conference on Reliability Infocom Technologies and Optimization Icrito 2016 Trends and Future Directions, 2016 Sensor nodes collect the data from the environment and send to sink. But attackers corrupt data while transmitting therefore data security is main concern of wireless sensor network (WSN). In proposed protocol, we decrease the passive attack on sink node by decreasing the traffic on sink node. The simulation results demonstrates the proposed method can each node will compress their data before sending to cluster head. After compressing, the packet size of node will decrease. This will decrease the traffic overload. In this compression technique, we reduce the size of packet by creating a code string of 0 and 1.
Improved energy efficiency semi static routing algorithm using sink mobility for WSNs Deepali, Padmavati 2014 Recent Advances in Engineering and Computational Sciences Raecs 2014, 2014 Sensor nodes work on batteries which are difficult to recharge and this makes energy proficiency as a main issue. One of the good solutions to this problem is clustering. Energy efficiency semi-static clustering (EESSC) protocol is an energy aware semi-static clustering protocol in which clustering is done based upon remaining energy. It considers all nodes in static mode. In this paper, a new routing protocol based on sink mobility i.e. improved EESSC (IEESSC) is proposed. This protocol is simulated and compared with EESSC and LEACH based on parameters stability period, network lifetime and middle node death (MND). Simulation results show that proposed protocol has better results than EESSC and LEACH.
