Bio health
Explore NeurEco's Impact in Medical Innovations.
NeurEco reduces development time of your medical applications:
-
Reducing the time needed to collect the data (less data)
-
Reducing the development time of the neural model
NeurEco enables more accurate predictions and provides enhanced explanations for those predictions.
Patient specific modeling
To support decision making
SPINNER European Project
Collaboration with ANSYS and AESCULAP
Make a real time copy of a computer intensive simulation tool
Real-time patient
specific simulation
Computer intensive simulation
Real time copy
Genomic classification
Non usual test case: the number of samples is almost 100 times smaller than the number of features (input variables)
Samples: 3450 = 2760 (learning data) + 690 (testing)
Features (embeddings): 300 000
Using the default setting of NeurEco
A Genetic example (classification)
Genome application
Confusion matrices
Confusion matrices obtained by Bengio [1] (left) and by NeurEco (right)
Classification error
Bengio & al. used several methods to solve the problem. The best classification error is 7.4 %.
The classification error of our method is 4.38 %
We reduced the error by 40 %
Bengio article: https://arxiv.org/pdf/1611.09340