Recognising body motion disorders at an early stage is critical for the timely diagnosis and effective treatment of various neurological and musculoskeletal conditions. Traditional diagnostic methods are often time-consuming and require expert intervention, which can delay the identification of these disorders. Machine learning technologies emerge with their powerful capacities to process large volumes of data, learn new data, and uncover subtle patterns and correlations. This paper introduces a novel vision-based abnormal action dataset (VBAAD), designed to leverage machine learning technologies for the recognition and analysis of body motion disorders, with the aim of empowering individuals to self-assess their own body health and detect potential problems early. The dataset, characterised by multiple views and multiple modalities, comprises 1,680 high quality videos capturing a wide range of abnormal and normal movement performance associated with predefined actions. Each video is recorded in controlled environments using multiple camera angles to ensure the front-half coverage of body movements. The dataset is meticulously curated, featuring diverse subjects with different degrees of motion disorders performing a set of predefined actions. The annotation of the dataset will facilitate precise training and evaluation of machine learning models. The dataset is collected using three Kinect-V2 cameras and one RealSense camera, which are positioned hemispherically to allow recording of the front half view of the participants simultaneously. Experimental results demonstrate the effectiveness of the proposed dataset in improving the accuracy of body motion disorder recognition systems. Comparative analysis is implemented to evaluate the performance of several state-of-the-art (SOTA) methods for multi-view multi-model vision-based body motion disorder recognition on VBAAD. After comparing the results of the models, it is indicated that VBAAD is a valuable resource for researchers and practitioners in the fields of computer vision, medical diagnostics, and rehabilitation. It provides a foundation for the development of advanced diagnostic tools and therapeutic interventions, ultimately contributing to more efficient and accessible healthcare solutions through the power of machine learning and empowering individuals with the capability for self-assessing their own body health.