Overview
Medical Informatics
Scrub
Datafly
Genomic identifiability
Patient-centered management
Surveillance
Selective-revelation
Risk assessment server
PrivaMix
Biometrics
Contactless capture
Policy and Law
Identifiability of de-identified data
HIPAA assessments
Privacy-preserving surveillance
Medical Informatics: Patient-centered management
[cite]
Work done with Andrew Halpert MD and Joan Waranoff of Blue Shield California.
Problem Statement: Demonstrate scientific experimentation with provably anonymous data: given dramatic increases in healthcare costs and the availability of effective interventions, develop automated methods that reduce costs by identifying cases for effective intervention.
Description: The first contribution of Dr. Sweeney and her colleagues was a multi-year cohort study that addressed whether patient-centered interventions can reduce utilization costs in complex patients over traditional case management without sacrificing life span. Subjects were 1.2mil HMO participants. They were able to show that patient-centered management can deliver more coordinated, cost effective care than traditional management, and can do so with high patient satisfaction and no adverse effect on survival. Overall costs were reduced by -26% (95% CI, 25-27%), which is a 2:1 return on investment after deducting costs of patient selection and management. This work used a social computing network of nurses to select patients for intervention; but in future work, they would like seek to develop an automated method for identifying patients for intervention based on fusing semantic and machine learning methods with third party data. The goal is to predict who will be an expensive patient without intervention, but who is also a good candidate for early intervention. Their work also demonstrates ways of conducting scientific experiments with provably anonymous data. 11
 |
Scientific Influence and Impact: The work of Dr. Sweeney and her colleagues seems to be the first to introduce an experimental design for comparing health outcomes of cohorts over time using provably anonymous data for analysis. The problem is important to insurance companies and the government who seek outside analysts to compute outcome measures. When using business or retrospective data, another problem is establishing like cohorts. Even though this work is recent, Hagan already reports other healthcare organizations (e.g., Healthnet, Alere, et al.) using variants of the experimental design.
Other Achievements: 12
- Privacy Leadership Award from Blue Cross Blue Shield Association (rarely awarded) for pioneering work on patient-centered management and using provably anonymous data.
Notes
| 11 |
One one hand, this work is not about privacy, but semantic and machine learning algorithms ("data detective work"). On
the other hand, the experimental design demonstrates health research with provably anonymous data.
|
| 12 | See quantitative assessments for more details. |
Related links: