Data Mining

Pattern Classification

We offer highly customized data mining services using machine learning tools such as regression models, unsupervised clustering methods, and supervised learning methods. These type of projects include but not limited to:

• Population subtype discovery
• Biomarker discovery
• Predictive model construction

Multi-Data Type Integration

Technology advancements have made it cost-effective to collect diverse types of high dimensional data. We offer network-based computational methods to combine multiple data types, such as clinical, questionnaires, image, and genomic data, for a set of samples to perform pattern recognition and model interactions.

• Identify sample similarity and community from multi-type data measurements
• Assess level of interaction between driver and respondents (i.e. gene expression driven by transcription factor binding)
• Isolate single sample contribution to network

Image Analysis using Deep Learning

Complex deep convolutional neural networks (CNNs), which are able “automatically” to capture predictive features in image data once appropriately trained, have garnered significant attention for their potential utility in biomedical applications with its accuracy. We provide deep learning neural network model construction service for quantitative image analysis. Applications include classification and regression modeling, as well as deep feature embedding for image data such as microscopy and histopathology.