November 10, 2015
Data
Data Description 1
- Data : "Physician and Other Supplier Data CY 2013"
- Part of the Obama Administration’s efforts to transform health care
- Prepared by The Centers for Medicare & Medicaid Services (CMS)
- Features
- 9287876 observations of 28 variables
- 909605 unique NPIs
- 90 provider types
- 5361 Healthcare Common Procedure Coding System(HCPCS) description
- Interesting features: information of values of claims submitted, medicare allowed and medicare payment (average and standard deviation)
Data Description 2
- Data : "Physician Referral Patterns - 2014 - 2015 365 day interval"
- Initially provided as a response to a Freedom of Information Act (FOIA) request
- Represent the number of referrals from one physician to another
More info https://downloads.cms.gov/FOIA/Data-Request-Physician-Referrals-Technical-Requirements.pdf
- Features
- 14080052 observations of 5 variables
- 750392 unique NPIs
- Containing referring physician NPI, referred physician NPI and referral count
Referral and clustering
Referred more or referring more?
- Directed graph is necessary for referral network. \(Referral\, Ratio\, = \frac{Referred}{Referring}\)
Community on referral network
- Similar referring pattern or similar referred pattern?
Clustering on referral network
Two types of community
- Referred community: community based on incoming edge, where providers with similar referred pattern are grouped.
- Referring community: community based on outgoing edge.
Q1: What's common feature within individual community?
- Type of provider? Radiation oncology is likely to refer patients to radiation oncology?
- Payment amount?
- Number of service provided?
- Number of referring or referred?
Q2: Why does long-distance referral happen?
- Pateints having two different houses (eg: summer house)
- Patients from rural area travelling to urban area for better medicare
- Well-known physician/specialty
Reference
Reference
K Rohe, T Qin, and B Yu. “Co-clustering for directed graphs; the stochastic co-blockmodel and a spectral algorithm Di-Sim.”
Kolaczyk, Eric D. Statistical Analysis of Network Data Methods and Models. [Online-Ausg.]. ed. New York, NY [u.a.: Springer, 2009. Print.