The rapid expansion of online social media platforms has revolutionized communication, enabling widespread interaction across diverse global audiences. However, this has also facilitated the rise of harmful behaviors, such as cyberbullying and toxic commentary. This research explores the multifaceted impact of cyberbullying on college students, focusing on its effects on academic performance, social life, mental health, and physical well-being. Data was collected using a structured questionnaire employing a 5-point Likert scale to assess the intensity of experiences and perceptions. Analyze and predict outcomes, deep learning models such as LSTM, Bi-LSTM, and GRU were implemented. These models were evaluated for performance across cyberbullying detection task, ensuring robustness and generalizability through metrics such as accuracy, precision, recall, F1-score, and AUC-ROC. The results indicate that the proposed models surpass baseline methods in both prediction accuracy and interpretability. This provides a comprehensive strategy to recognizing and tackling cyberbullying in higher education settings.