Multimodal Data Analysis
We uncover the relationship between inspection results, medical history and the social (and natural) environment by mining massive multi-modal data such as text and image. We uncover the disease model and the hidden information modal among groups, establish the predictive analysis modal, further explore the pattern of disease distribution and evolution, define the risk factors, and predict the trend of disease, providing decision makers with sustainable and reliable information to make disease monitoring and public health policies.
We are persistent in working with top-ranking general hospitals, specialized hospitals and medical health data centers worldwide, in order to obtain high-quality clinical medical data for the sake of modelling. Data cleansing, text recognition, and natural semantic parsing are carried out on the basis of massive authentic desensitized clinical data; Deep-learning models are trained for the analysis of relevance and contribution of multiple dimensional parameters and morbidity among patients. We unearth the potential correlativity among patients and patients’ behaviors, medical history, and genes, and build the prediction models.