You were given a data set consisting of 500 variables related to hospital emergency department (ED) patients and tasked with identifying groups of patients most likely to revisit the ED within 72 hours of their initial visit. After several weeks of analysis, you presented a model based on k-means clustering with a sensitivity of 25% and specificity of 75% (a remarkable achievement). However, despite the health system's best efforts, attempts to operationalize this model have all failed. Though the model is precise at identifying patients likely to revisit the ED, there does not seem to be any intervention that can be effectively deployed to avoid these revisits. Name at least two possible reasons why attempts to intervene with this cluster of patients have failed. Do you think this data mining effort was successful? Why or why not? What do you think your next steps should be?

Running head: CLUSTERING  1  Clustering Name Institution  CLUSTERING  2 Clustering  Approximately 3% of patients discharged from the emergency department

CLUSTERING. 1. Clustering Name Institution. CLUSTERING.

End of preview