Hua Fang, Associate Professor of Computer & Information Science at UMass Dartmouth’s College of Engineering, has received a patent for a "system and methods for trajectory pattern recognition" development. The invention improves the accuracy and the stability of identified patterns, generally the structure of high-dimensional data with missing values. “It dynamically displays and clusters correlated high-dimensional, longitudinal data with missing values, potentially useful for real-time data,” Fang says. It also involves pattern recognition techniques used for analyzing heterogeneity effects of behavioral interventions.
“The improved accuracy and stability of longitudinal trajectory patterns can be used for a variety of analytic and assessment purposes, ranging from biomedical, real-time sensor data analyses, risk prediction, policy implementation, consumer behavior, business intelligence to security and fraud detection including image authentication,” Fang explains. “It could also cover all types of trajectory pattern recognition. Traffic trajectories of autonomous vehicles may be another example, and potentially trajectory recovery with missing values.”
The invention is also applicable to a wide variety of applications involving longitudinal observational studies and clinical trials, including typical randomized clinical trials and pragmatic trials. Fang explains that patterns and approaches may be used to characterize complex dietary response behaviors and untangle the effect of dietary patterns on various outcomes, such as body weight, incidence of diabetes, and atherosclerotic cardiovascular disease events including myocardial infarction, stroke, and cardiovascular disease death.
Furthermore, the parameter trajectory description can assist in real-time biosensor data analytics for detecting episodes in behavioral science and drug abuse. The development is applicable to infectious and immunology data, detecting different and correct patterns for measles virus data, and may be utilized in health risks and policy studies such as those involving Medicare and Medicaid data.
It may also be used for consumer behavior data that pertains to investment and credit risk by evaluating the effectiveness of contrarians' investment in preferred stocks using big data, shopping, and marketing behavior in e-commerce. This invention was made with government support and its related grant was awarded by the National Institute of Health.