Coursera Data Products

According to the Skin Cancer Foundation, “Melanoma” is the most dangerous form of skin cancer. These cancerous growths develop when unrepaired DNA damage to skin cells (most often caused by ultraviolet radiation from sunshine or tanning beds) triggers mutations (genetic defects) that lead the skin cells to multiply rapidly and form malignant tumors. Given a family history of skin cancer and my own experience in the health care industry, this analysis uses a publicly available dataset (CMS.gov) to look at the performance of health care provider groups that report on the disease.

This web application uses the following data set:

Physician Compare 2016 Group Public Reporting https://data.medicare.gov/Physician-Compare/Physician-Compare-2016-Group-Public-Reporting/dyub-7qq9

• This file contains Physician Quality Reporting System (PQRS) and non-PQRS Qualified Clinical Data Registry (QCDR) measure performance rates reported by groups. This dataset was updated on May 17, 2018.

The web application may be found here: https://marc905.shinyapps.io/Prescriber_Group/

The Prescriber Group data set was reduced to a subset of the performance metrics and providers that focus just on Melanoma. Linear models were then created to explore the relationship between the category of “Melanoma: Coordination of Care” and other Melanoma-related metrics.

• Benchmark score being predicted: Melanoma: Coordination of Care
• Model 1 Input: Melanoma Reporting
• Model 2 Input: Model 1 + Melanoma: Continuity of Care - Recall System
• Model 3 Input: Melanoma: Overutilization of Imaging Studies in Melanoma

• Model 1: Sparse population of provider groups that report on both Coordination of Care and Melanoma Reporting metrics
• Model 2: Recall System metrics appear somewhat correlated with outcome
• Model 3: Metric generally scored at 100%

Unfortunately, all three models returned a R squared value between 0.2 and 0.25. Therefore, I'd conclude that none of the models explained a great deal about a given care provider organization expected Coordination of Care proficiencies. At the same time, I've learned a great deal about the data sets that CMS makes publicly accessible, the Shiny platform and creating presentations through R.

summary(linear_model1)$coef

                                          Estimate Std. Error  t value
(Intercept)                             71.4465114 8.64141684 8.267916
Measure.Performance.Rate.PQRS_GRP_397_1  0.2277743 0.09613261 2.369376
                                            Pr(>|t|)
(Intercept)                             6.981272e-08
Measure.Performance.Rate.PQRS_GRP_397_1 2.799270e-02