Some basic health information of the human body, such as age, weight, gender, etc., can be obtained through gait analysis. We propose a unified gait health analysis framework for multimodal and multiattribute prediction tasks. In the proposed framework, we integrate silhouette and optical flow modalities and employ a gait feature extractor to extract robust gait features. We also design a Mixture-of-Experts (MoE) fusion module to dynamically combine complementary static and dynamic information, enabling a shared representation for predicting multiple biometric and health-related attributes. The extensive experiments on the Health&Gait dataset show our proposed models outperform baseline single-modality methods across various prediction tasks, validating the effectiveness of multimodal gait analysis in practical health assessment scenarios.