Thanks for this app. I have been experimenting with this and learning the paper (related papers) to understand how this approach can help in labeling
I understand that APHRODITE creates cases (mention of keywords) and controls (no mention of keywords).
Let's say I have 3000 patients. APHRODITE identifies 1800 as cases and 800 as controls and we extract data for these cases and control using get patient data function and build a model using them. Based on my experiments, I believe this 2600 will be split into train and test.
Later if I want to evaluate my model on the remaining 400 patients (unseen patients), do I have to prepare the patient id list (getting everything other than 2600 patients) and extract the data again using get patient data function? Am I right to understand this? I was experimenting like this. But curious to know whether APHRODITE has a function which will create an evaluation cohort (consisting of 400 patients)?
Hello,
Thanks for this app. I have been experimenting with this and learning the paper (related papers) to understand how this approach can help in labeling
I understand that APHRODITE creates cases (mention of keywords) and controls (no mention of keywords).
Let's say I have 3000 patients. APHRODITE identifies 1800 as cases and 800 as controls and we extract data for these cases and control using get patient data function and build a model using them. Based on my experiments, I believe this 2600 will be split into train and test.
Later if I want to evaluate my model on the remaining 400 patients (unseen patients), do I have to prepare the patient id list (getting everything other than 2600 patients) and extract the data again using
get patient data
function? Am I right to understand this? I was experimenting like this. But curious to know whether APHRODITE has a function which will create anevaluation cohort
(consisting of 400 patients)?