Study groups to Received actions labeled as reciprocal determine - - PowerPoint PPT Presentation

Study groups to determine if board

• rder narratives

provide additional insight beyond the Conclusions of Law.

─Received their first-ever action from a board ─Represent 45% of disciplined physicians in 2017 ─Have 1.3 board orders and 2.5 actions per physician

Group 1

─Received actions labeled as reciprocal ─Represent 13% of disciplined physicians in 2017 ─Have 4.6 board orders and 7.7 actions per physician

Group 2

─Received multiple actions all from the same board ─Represent 26% of disciplined physicians in 2017 ─Have 3.3 board orders and 5.8 actions per physician

Group 3

─Received multiple actions from different boards ─Represent 45% of disciplined physicians in 2017 ─Have 6.1 board orders and 9.8 actions per physician

Group 4

 Board order narratives usually give more complete information than ‘Conclusions of Law’ therefore it is appropriate to move ahead with a categorization project.  Documents received from boards vary in quality, format and detail. This makes automation non-feasible at this time.  Challenges with reciprocal actions remain. For statistical purposes, if an action is known to be reciprocal, it will be classified as such, regardless of associated narrative.  ‘Not applicable’ basis code will be classified as ‘No reason given.’

Observations

Results

Bayesian method yielded the highest accuracy for entire documents. Accuracy on three classes: averages 40%, up to 60%

• n certain training sets.

fastText provides the highest accuracy for slices which could be used to assist a human evaluator.

These findings seem low – but are statistically significant compared to random guessing. These models were trained

• n limited data pools and could improve with more data.