Healthcare Data Quality
Health Services Analytics
Technology Adoption
AI/ML in Healthcare
My research examines how healthcare data and information systems can be made more reliable, interpretable, and actionable
for clinicians, administrators, and caregivers. I focus on applied, real-world datasets and reproducible analytics,
with attention to implementation constraints in public-sector and care delivery settings.
Research Areas
Applied AI/ML for Healthcare Decision Support
When AI/ML is used in healthcare, it must be reliable, explainable where appropriate, and aligned with realistic
deployment constraints. I focus on applied modeling for healthcare tasks, emphasizing evaluation, interpretability,
and implementation feasibility.
Representative directions
- Explainable modeling and structured evaluation in clinical/healthcare datasets.
- Bridging performance with deployment constraints (runtime, memory, and system fit).
- Use of AI methods as decision support rather than opaque automation.
Health Services Analytics and Performance Measurement
I study how provider performance and patient outcomes vary across geography, organizational characteristics,
and population risk. This includes performance measurement in home healthcare and other service settings, with
an emphasis on interpretable models that support stakeholders.
Representative directions
- Performance comparisons across rural–urban classifications and service environments.
- Outcome-focused analytics (e.g., discharge to community, utilization patterns).
- Risk score determinants and relationships to access and provider availability.
Typical data
- Home Health Compare (HHC) and related public quality datasets
- CMS risk score frameworks (e.g., HCC-related analyses)
Healthcare Data Quality and Governance
Healthcare analytics often fails not because the model is weak, but because the data is incomplete, inconsistent,
or mis-specified. My work explores practical methods to detect data defects, quantify their impact, and support
remediation workflows that improve downstream reporting and decision-making.
Representative directions
- Defect detection in administrative datasets (e.g., eligibility, claims, and utilization fields).
- Quality metrics and monitoring pipelines for longitudinal healthcare data.
- Tooling that supports continuous improvement across teams and agencies.
Typical data
- Medicaid Management Information System (MMIS) data
- All-Payer Claims Data (APCD)
Technology Adoption by Providers and Caregivers
Many health IT innovations fail at the adoption layer. I investigate sociotechnical barriers and enablers of
technology use among caregivers and healthcare stakeholders, including digital literacy, trust, perceived value,
and workflow fit.
Representative directions
- Adoption constraints and usability challenges in caregiver-facing technologies.
- Socio-demographic drivers of access and service usage in digital health contexts.
- Practical recommendations for implementing systems in resource-constrained settings.
Methods & Data
Analytical Approach
- Data management: cleaning, normalization, validation checks, reproducible ETL pipelines.
- Statistics & modeling: interpretable regression/GLM, machine learning where suitable, robust evaluation.
- Visualization: stakeholder-facing dashboards, performance comparisons, and clear reporting.
- Software systems: tool development and evaluation for operational public-sector environments.
Representative Data Sources
- Medicaid Management Information System (MMIS) data
- All-Payer Claims Data (APCD)
- Home Health Compare (HHC) and related public CMS datasets