|
@HerffC | |||||
|
And a new #tweeprint out with @jeroenhabets1, @DigNeurosurgeon Mark Janssen and others:
"Machine learning prediction of motor response after deep brain stimulation in Parkinson's disease"
medrxiv.org/content/10.110…
|
||||||
|
||||||
|
Christian Herff
@HerffC
|
7. lis |
|
In this study, we trained a logistic regression to predict the motor outcome of Deep Brain Stimulation one year after surgery from preoperative variables. pic.twitter.com/a3e8lov0Nl
|
||
|
|
||
|
Christian Herff
@HerffC
|
7. lis |
|
For this, we divided our 90 patients into weak and strong responders: pic.twitter.com/giRRfOKqLY
|
||
|
|
||
|
Christian Herff
@HerffC
|
7. lis |
|
Our prediction model achieves very good quality (AUC = 0.88) in a cross-validation. By analyzing the odd-ratios in the logistic regression, we can see which preoperative variables have the highest impact on the prediction. pic.twitter.com/iH4eS6kCoM
|
||
|
|
||
|
Christian Herff
@HerffC
|
7. lis |
|
While old age at PD onset generally predicted weaker responders, a high UPDRS IV score predicted better motor outcome. Interestingly, we didn't see a large predictive value of the Levodopa response.
|
||
|
|
||