Evaluating Mental Health Awareness on TikTok and Twitter with Sentiment Analysis and XGBoost

The impact of social networks on mental health awareness has become increasingly significant as technology has become more integrated into our society. As it grows, it becomes increasingly crucial to understand the factors that build engagement with mental health-related content. Factors include post length, post sentiment, hashtags, and chosen platforms. Carefully analyzing which factors are the most influential can help people create meaningful posts with high engagement rates to help raise awareness of mental health. Using regression and classifying models, we categorize posts into high, medium, and low engagement levels, and aim to predict engagement rates based on the aforementioned factors. The most effective models from our experiments are the Gradient Boost and XGBoost models, which identified positive sentiment and specific hashtags to be the most influential. They demonstrated the highest performance with our XGBoost model having an overall accuracy and F1-score of 0.81, and our Gradient Boost model having an overall accuracy and F1-Score of 0.78. Both of these models excelled in classifying engagement levels in to low, medium, and high categories on TikTok. These insights provide us with recommendations on how to create content that is efficient in reaching people and raising awareness of mental health.