ADHD Validation (CCHMC)

Algorithm Type: 
Case, Control
Institution: 
Cincinnati Children's Hospital Medical Center
Implementation Details: 

We implemented the ADHD algorithm and performed a chart review of 46 cases and 40 controls. The task was split between two annotators, one a ADHD physician, and another a genetics expert. A significant amount unknowns prevail in these cases and controls due to insufficient documentation. A question remains whether these unknowns should be considered yes or no. In the implementation results, we considered all unknowns to be 'yes' (true positive) for either case or controls. 

If unknowns are marked as no (false positive), then the case PPV = 67.4% and the control PPV = 62.5%.

Phenotype: 
ADHD phenotype algorithm
Dictionaries
Implementations/Datasets/
Cases (0) Actual Class (Expectation)
Predicted Class
(Observation)
True Positive: 
41.00
Correct result
False Positive: 
5.00
Unexpected result
Case Positive Predictive Value:
 89.13%
Control (0) Actual Class (Expectation)
Predicted Class
(Observation)
True Positive: 
38.00
Correct result
False Positive: 
2.00
Unexpected result
Control Positive Predictive Value:
 95%