Generative AI-based Healthcare Recommender System

Personalized healthcare recommendations remain challenging due to diverse patient data, including medical history and lifestyle habits. Traditional systems struggle to provide real-time, personalized recommendations, leading to ineffective treatment. This research improves healthcare recommendation systems (HRS) using generative AI techniques, specifically Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), to enhance personalization, accuracy and adaptability. This study explores synthetic data generation to address data sparsity and cold-start problems while maintaining privacy. Exploratory Data Analysis (EDA) and preprocessing methods like feature engineering, identification of missing data, normalization and outlier detection are part of the research methodology. Interpretability is enhanced by data visualization using boxplots, histograms and heatmaps. Although complete GAN and VAE implementation was not possible due to computational limitations, baseline assessments created a fundamental framework. According to preliminary findings, generative models can fill in the gaps in customisation. Potential improvements in prediction performance are shown by evaluation criteria including Root Mean Square Error (RMSE), accuracy and precision. Despite its drawbacks, this research shows that integrating Variational Autoencoders (VAEs) into HRS is viable for improved scalability and flexibility.