Building an Open Health Data Analytics Platform: a Case Study Examining Relationships and Trends in Seniority and...

With a movement toward an open data-driven approach, governments worldwide are releasing public data in an effort to increase transparency. Despite the availability of this data, many factors make it difficult for anyone to extract knowledge from it. The relationship between seniority and performance is a controversial issue in many fields; using an open and reproducible framework, we investigate the relationship in open healthcare data. Using data from the Center for Medicare and Medicaid Services covering 895,000 practitioners and 3000 hospitals, weak but statistically significant correlations between graduation year, a proxy for seniority, and the hospital value-based performance score were found in 29 of 74 specialties (Spearman rank correlation values < 0.164, p value < 0.05). This result represents 7% of US healthcare practitioners and over 75% of medical practitioners in several specialties. With 5 years of data (2009-2014) from the New York Statewide Planning and Research Cooperative System (SPARCS), we found weak but statistically significant correlations between graduation year and cardiac surgery outcome measures (Spearman rank correlation value - 0.096, p value < 0.0005). An unsupervised clustering K-means-based algorithm for finding outliers was also applied to these datasets. It captured a unique trend in the number of nurse practitioners which was increasing rapidly since 2010. It also revealed consistencies in practitioner placement throughout hospitals. Our findings suggest that the training of healthcare professionals appears to be robust and positions them for long-lasting and consistent careers across the majority of specialties.