This Julia toolbox allows predicting clozapine (CLO) and norclozapine (NCLO) blood concentrations.
Individual recommended safe dose range can also be predicted:
The models are actively developed and we expect their accuracy to improve.
[ Info: Creating classifier model
Initial cross-entropy: 0.222
[ Info: Optimizing: n_hidden
Progress: 100%|████████████████████| Time: 0:01:56
[ Info: Optimizing: dropout
Progress: 100%|████████████████████| Time: 0:01:22
[ Info: Optimizing: epochs
Progress: 100%|████████████████████| Time: 0:06:02
[ Info: Optimizing: batch_size
Progress: 100%|████████████████████| Time: 0:00:02
[ Info: Optimizing: η
Progress: 100%|████████████████████| Time: 0:00:19
[ Info: Optimizing: λ
Progress: 100%|████████████████████| Time: 0:01:45
[ Info: Optimizing: α
Progress: 100%|████████████████████| Time: 0:06:37
Final cross-entropy: 0.5052
Optimized parameters:
n_hidden: 32
dropout: 0.01
η: 0.01
epochs: 500
batch_size: 2
λ: 0.0
α: 0.001
[ Info: Classifier accuracy
Training:
cross-entropy: 0.0
log-loss: 0.0
AUC: 1.0
misclassification rate: 0.0
accuracy: 1.0
Confusion matrix:
sensitivity (TPR): 1.0
specificity (TNR): 1.0
group
norm high
┌──────┬──────┐
norm │ 74 │ 0 │
prediction ├──────┼──────┤
high │ 0 │ 21 │
└──────┴──────┘
Validating:
cross-entropy: 0.58
log-loss: 0.58
AUC: 0.98
misclassification rate: 0.07
accuracy: 0.93
Confusion matrix:
sensitivity (TPR): 0.9
specificity (TNR): 0.94
group
norm high
┌──────┬──────┐
norm │ 29 │ 1 │
prediction ├──────┼──────┤
high │ 2 │ 9 │
└──────┴──────┘
[ Info: Creating regressor model: clozapine
Initial RMSE: 129.7905
[ Info: Optimizing: n_hidden
Progress: 100%|████████████████████| Time: 0:04:11
[ Info: Optimizing: dropout
Progress: 100%|████████████████████| Time: 0:03:14
[ Info: Optimizing: epochs
Progress: 100%|████████████████████| Time: 0:12:58
[ Info: Optimizing: batch_size
Progress: 100%|████████████████████| Time: 0:00:18
[ Info: Optimizing: η
[ Info: Optimizing: λ
Progress: 100%|████████████████████| Time: 0:03:14
[ Info: Optimizing: α
Progress: 100%|████████████████████| Time: 0:30:43
Final RMSE: 118.4395
Optimized parameters:
n_hidden: 64
dropout: 0.52
η: 0.01
epochs: 1000
batch_size: 2
λ: 0.1
α: 0.0
[ Info: Regressor accuracy: clozapine
Training:
R²: 0.96
RMSE: 66.61
Validating:
R²: 0.85
RMSE: 132.57
[ Info: Creating regressor model: norclozapine
Initial RMSE: 84.6936
[ Info: Optimizing: n_hidden
Progress: 100%|████████████████████| Time: 0:04:13
[ Info: Optimizing: dropout
Progress: 100%|████████████████████| Time: 0:03:26
[ Info: Optimizing: epochs
Progress: 100%|████████████████████| Time: 0:14:05
[ Info: Optimizing: batch_size
Progress: 100%|████████████████████| Time: 0:01:28
[ Info: Optimizing: η
[ Info: Optimizing: lambda
Progress: 100%|████████████████████| Time: 0:13:49
[ Info: Optimizing: α
Progress: 100%|████████████████████| Time: 2:07:02
Final RMSE: 73.4603
Optimized parameters:
n_hidden: 128
dropout: 0.54
η: 0.01
epochs: 8400
batch_size: 4
λ: 0.1
α: 0.226
[ Info: Regressor accuracy: norclozapine
Training:
R²: 0.97
RMSE: 30.0
Validating:
R²: 0.83
RMSE: 69.92
[ Info: Training final model
Classifier accuracy:
log-loss: 0.01
AUC: 1.0
misclassification rate: 0.01
accuracy: 0.99
Confusion matrix:
sensitivity (TPR): 1.0
specificity (TNR): 0.99
group
norm high
┌──────┬──────┐
norm │ 104 │ 0 │
prediction ├──────┼──────┤
high │ 1 │ 31 │
└──────┴──────┘
Predicting: clozapine
Regressor accuracy
R²: 0.97
RMSE: 63.54
Predicting: norclozapine
Regressor accuracy
R²: 0.97
RMSE: 28.78
Regressor:
Subject ID: 1 CLO level: 240.5 prediction: 74.7 RMSE: 165.8
Subject ID: 1 NCLO level: 90.7 prediction: 110.3 RMSE: 19.6
Subject ID: 2 CLO level: 292.4 prediction: 793.8 RMSE: 501.4
Subject ID: 2 NCLO level: 283.8 prediction: 343.0 RMSE: 59.2
Subject ID: 3 CLO level: 390.5 prediction: 593.8 RMSE: 203.3
Subject ID: 3 NCLO level: 162.5 prediction: 250.1 RMSE: 87.6
Subject ID: 4 CLO level: 586.1 prediction: 771.9 RMSE: 185.8
Subject ID: 4 NCLO level: 189.4 prediction: 321.7 RMSE: 132.3
Regressor accuracy:
Predicting: CLOZAPINE
R²: -4.09
RMSE: 297.8
Predicting: NORCLOZAPINE
R²: -0.52
RMSE: 85.24
Classifier:
Subject ID: 1 group: NORM prediction: NORM, prob = 1.0 adj. prediction: NORM, prob = 1.0
Subject ID: 2 group: NORM prediction: HIGH, prob = 0.69 adj. prediction: NORM, prob = 0.61
Subject ID: 3 group: NORM prediction: HIGH, prob = 0.63 adj. prediction: NORM, prob = 0.67
Subject ID: 4 group: HIGH prediction: HIGH, prob = 0.92 adj. prediction: HIGH, prob = 0.62
Classifier accuracy:
log_loss: 0.56
AUC: 1.0
misclassification rate: 0.5
accuracy: 0.5
Confusion matrix:
sensitivity (TP): 1.0
specificity (TP): 0.33
group
norm high
┌──────┬──────┐
norm │ 1 │ 0 │
prediction ├──────┼──────┤
high │ 2 │ 1 │
└──────┴──────┘
Adjusted classifier accuracy:
log_loss: 0.35
AUC: 1.0
misclassification rate: 0.0
accuracy: 1.0
Confusion matrix:
sensitivity (TP): 1.0
specificity (TP): 1.0
group
norm high
┌──────┬──────┐
norm │ 3 │ 0 │
prediction ├──────┼──────┤
high │ 0 │ 1 │
└──────┴──────┘
Clone this repository, go to its folder and run:
julia src/server.jlNext, go to the local website at http:https://localhost:8080, enter patient's data and click the "PREDICT" button.
(!) Adjusted clozapine level is the classifier prediction modified by predicted concentration.
Toxic clozapine level has been defined as > 550 ng/mL, recommended therapeutic concentration has been defined as > 220 ng/mL [source: 10.1192/bjp.2023.27].
If you use this tool, please acknowledge us by citing our paper.
Below is the list of contributors and their affiliations.
This software is licensed under The 2-Clause BSD License.
DISCLAIMER: THIS TOOL HAS THE RESEARCH USE ONLY (RUO) STATUS
This tool and all associated information, including but not limited to, text, graphics, images and other material contained on this website, has the Research Use Only (RUO) status. It is intended for scientific research only. It must not be used for diagnostic or medical purposes.
DISCLAIMER: THIS WEBSITE DOES NOT PROVIDE MEDICAL ADVICE
This tool and all associated information, including but not limited to, text, graphics, images and other material contained on this website are for informational purposes only. No material on this site is intended to be a substitute for professional medical advice, diagnosis or treatment. Always seek the advice of your physician or other qualified health care provider with any questions you may have regarding a medical condition or treatment and before undertaking a new health care regimen, and never disregard professional medical advice or delay in seeking it because of something you have read on this website.