Since the Dartmouth hospital service areas (HSAs) were proposed three decades ago, there has been a large body of work using the unit in examining the geographic variation in health care in the U.S. for evaluating health care system performance and informing health policy. However, many studies question the replicability and reliability of the Dartmouth HSAs in meeting the challenges of ever-changing and a diverse set of health care services. This research develops a reproducible, automated, and efficient GIS tool to implement Dartmouth method for defining HSAs. Moreover, the research adapts two popular network community detection methods to account for spatial constraints for defining HSAs that are scale flexible and optimize an important property such as maximum service flows within HSAs. A case study based on the state inpatient database in Florida from the Healthcare Cost and Utilization Project is used to evaluate the efficiency and effectiveness of the methods. The study represents a major step toward developing HSA delineation methods that are computationally efficient, adaptable for various scales (from a local region to as large as a national market), and automated without a steep learning curve for public health professionals.
Keyphrases
- healthcare
- mental health
- deep learning
- machine learning
- high throughput
- quality improvement
- health information
- randomized controlled trial
- public health
- systematic review
- real time pcr
- loop mediated isothermal amplification
- palliative care
- acute care
- adverse drug
- label free
- risk assessment
- social media
- affordable care act