1. Stochastic models & service mechanism for care delivery w. health-IT
    Fundamental methodological challenges in studying healthcare systems include system variability, uncertainty, heterogeneity, and providers’ and patients’ strategic behaviors. The HEALTH-Engine Lab aims to develop rigorous and computationally efficient quantitative methods to address the challenges, and apply them in strategic and operational-level health-IT (e.g., electronic service) implementation. The research team will build stochastic models to measure ambulatory care delivery system performance featuring e-visit and e-consult, and design service and incentive mechanisms for coordinating primary care physicians and specialists and engaging patients to adopt and adapt to web-based care delivery modality.
  2. Simulation and optimization for health policy & operations
    The HEALTH-Engine Lab uses discrete-event, systems dynamics, agent-based, and multi-method simulations to model complexities of specific healthcare environments. Optimization models (stochastic programming, game theory, and Markov decision processes) are valuable tools to address decision-making under uncertainty in the healthcare system. The simulation and optimization models are integrated to derive solutions for healthcare operations (out-patient appointment scheduling, inpatient discharge services) and health policy (readmission reduction programs, prescription drug programs).
  3. Data-driven decision-making for diagnosis & treatment​
    Traditional data-driven decision-making theories and tools apply to problems under the increasingly restrictive assumptions: all the samples should be stored in one location and the sample size must increase polynomially with the growth of the problem dimensions. With the surging need for more comprehensive and higher-granule models in the contemporary medical applications, the dimensionality of data-driven medical decision problems is rapidly inflating, and so does the required size of locally accessible datasets. As a result, the data availability and the storage capacity repetitively run short. In view of the existing challenges, The HEALTH-Engine Lab is studying new high-dimensional learning tools and distributed data-driven optimization paradigms that are substantially less contingent on data abundance and are compatible with distributed medical datasets.