Social media-based machine learning method for predicting symptoms in people with post-traumatic stress disorder with cancer

One of the numerous mental health issues that cancer survivors often deal with is post-traumatic stress disorder (PTSD). Traditional approaches to identifying PTSD symptoms mostly rely on professional evaluations and self-reported data, which can be time-consuming and have a limited sample size. Due to the platforms’ rapid growth, social media users are sharing more intimate experiences and emotional states online, making them a valuable source of data for mental health research. This study proposes a social media-based machine learning framework that uses artificial neural networks (ANN) to detect PTSD symptoms in cancer survivors. Social media communications collected from several platforms were preprocessed using natural language processing techniques such as tokenization, stop-word removal, and Word2Vec word embedding. An Convolutional Neural Network (CNN) based model that could categorize postings as either suggestive or non-indicative of PTSD symptoms was trained using the retrieved characteristics. The suggested approach successfully detects PTSD-related information from social media posts, according to experimental data. The Convolutional Neural Network (CNN) based-based method showed good predictive performance with an overall classification accuracy of 95%, precision of 0.93, recall of 0.95, and F1-score of 0.94. These findings demonstrate the possibility of integrating deep learning methods with social media data for the early identification of PTSD symptoms in cancer survivors. By permitting rapid mental health intervention and large-scale monitoring of psychological discomfort, the suggested framework might be a helpful tool for healthcare workers.