In this study, we aim to explore whether rephrasing enhances the performance of advanced question answering systems in terms of accuracy. Initially, we analyze the effectiveness of Bidirectional Encoder Representations from Transformers (BERT)—a widely used pre-trained language model—when presented with rephrased questions and documents containing distracting details. Our findings reveal that BERT exhibits excessive sensitivity to such instances, prompting us to investigate potential strategies for mitigating this sensitivity. To address the observed decline in performance, we propose a refinement approach for BERT, incorporating techniques such as data augmentation and multitask learning. Specifically, given a question and its context, we generate a series of paraphrases through back-translation. Alongside minimizing the loss between the predicted answer distributions for the original questions—we also minimize the supervised loss associated with the augmented questions. Furthermore, we introduce an auxiliary objective aimed at minimizing the unsupervised loss between the answer distributions of the original and augmented questions. Notably, this auxiliary loss is unsupervised, as it does not directly rely on labels corresponding to the augmented examples. To compute this unsupervised loss, we employ measures such as symmetric Kullback-Leibler divergence and the Jensen-Shannon distance, serving as regularization techniques for the model. The findings revealed that the supervised model exhibited superior performance compared to both the original and our proposed model. However, our proposed model demonstrated a nearly 2% enhancement in both Exact Match (EM) and F1 scores when tested on a paraphrased development set that we formulated. This suggests the potential effectiveness of our proposed approach in enhancing model accuracy and performance.

Immunotherapy, particularly Immune Checkpoint Blockade (ICB), has demonstrated significant efficacy in treating melanoma in recent years. However, accurately predicting treatment success and avoiding ineffective therapies remains an unresolved challenge. Therefore, this study aims to develop statistical models utilizing neural networks to forecast the effectiveness of immune checkpoint therapies for melanoma patients. Our models primarily rely on Artificial Neural Networks (ANN) to anticipate both Overall Survival (OS) and Progression-Free Survival (PFS) among melanoma patients undergoing anti-CTLA4 and anti-PD1/anti-PDL1 therapy. We incorporate gene expression data, measured in Transcripts per Million (TPM), derived from bulk tumor RNA-sequencing datasets. Additionally, clinical variables such as gender, age, and treatment type are factored into our analysis. The ANN underwent optimization to attain the highest feasible precision in anticipating the predetermined survival outcome. Issues stemming from high-dimensional data, such as overfitting, were tackled through regularization and feature selection methods. Consequently, the ANN-based model incorporating feature selection exhibited the capacity to forecast survival (PFS) in response to ICB therapy with a maximal precision of 86%. Conversely, the ANN lacking feature selection but incorporating regularization achieved accuracies of up to 72% for PFS and 71% for OS, correspondingly. In order to confront the challenge posed by limited patient samples and to assess replicability, the model underwent training and validation based on the amalgamation of all five datasets. However, this amalgamation failed to enhance predictive performance, necessitating further investigations.