Kai Sun, Ph.D.

Kai Sun, Ph.D.

Assistant Professor of Information Systems, UMBC | Adjunct Assistant Professor, Long School of Medicine, UT Health San Antonio

Research

RESEARCH INTERESTS

My primary research interests center around optimal and equitable resource planning in service systems leveraging large-scale datasets via the development and application of
(a) Optimization under uncertainty
(b) Multi-objective optimization
(c) Graph-based models and algorithms
(d) Data analytics and machine learning

The examples of the applications includes:
(1) In healthcare: operating rooms and intensive care units management, data-driven patient and clinician scheduling;
(2) In product development and manufacturing: process planning and scheduling, product lifecycle management (PLM) systems, product-service manufacturing systems.

PROJECTS

Wellbeing-Aware Scheduling for Emergency Medicine and Intensive Care, at UMBC in collaboration with UMB;

  • Developing a wellbeing-aware scheduling platform that combines electronic health record (EHR) data, clinician schedules, validated surveys, and organizational policies to support sustainable workforce planning.
  • Building clinician-centered knowledge graphs and graph databases to represent clinicians, shifts, clinical activities, teamwork patterns, workload, and wellbeing in a unified and explainable framework.
  • Integrating interpretable machine learning and graph-based models with robust, multi-objective optimization to generate schedules that balance coverage, workload fairness, fatigue risk, clinician wellbeing, and duty-hour requirements.

Machine Learning-Guided Clinician Wellbeing Metrics and Scheduling Tools, at UMBC in collaboration with UT Health San Antonio and UMB;

  • Developing clinician workload and wellbeing metrics from EHR activity, work schedules, human resources data, and surveys, using the NASA Task Load Index and Mayo Clinic Well-Being Index as validation baselines.
  • Applying interpretable machine learning and causal modeling to identify how clinician characteristics, shift attributes, patient complexity, and recent schedules influence workload, burnout risk, and wellbeing.
  • Developing an open-source predict-then-optimize scheduling tool that incorporates individual preferences, fair workload distribution, operational efficiency, and uncertainty into multi-objective shift assignments.

Optimal Scheduling of Physicians and Certified Registered Nurses, at UTSA and UTHealthSA;

  • Developed data-driven scheduling and analytics framework with novel mixed- integer multi-objective programming formulation and solution methods to support clinical operations and strategic decisions.
  • Embedded in the Department of Anesthesiology, working hand-in-hand with the front-line clinicians and administrators from UTHealthSA.
  • The developed framework has been implemented and deployed for anesthesiology scheduling and operational analysis with more than 200 providers. This improved physician schedule satisfaction from 3.13/5 to 3.47/5, and relieved at least 82% of scheduling burdens on departmental leaders.

Development and Application of Data Analytics for Smart PLM systems, at SU and collaborated with two local companies, i.e., Filtertech, Inc. and UsPLM, Inc.;

  • Developed a novel algorithm framework for the maximum weighted independent set (MWIS) problem, exact and improved estimation algorithms were created using the framework.
  • Developed a novel graph-based formulation approach for process planning and scheduling problems with its solution methods using the concept of MWIS.
  • Built product design libraries (50+ products and 500+ components) for small kitchen appliances and unmanned aerial vehicles (UAVs) [Product Design Repository].
  • Developed visual analytics, product configuration, resource allocation, and process planning and scheduling modules for the Smart PLM system, and deployed these tools for the engineering design department at Filtertech, Inc. [Engineering Analytics Toolset].
  • Developed and deployed digital twin and flight mission configuration services for the drone fleet management solution at UsPLM, Inc., Key contributor to win Genius NY 2018 investment competition, $250,000 (external, NY State and The Tech Garden; Featured in the front page of The Tech Garden) [Demo1] [Demo2] [Demo3] [Demo4];
  • Development of Educational and Training Materials for Unmanned Aerial Systems, wrote the proposal for the Gryphon Sensors grant, $20,000 (external).