The RaDiCo <scp>AC</scp>‐<scp>OEIL</scp> : a French rare disease cohort dedicated to ocular developmental anomalies in children

Calvas, Patrick; Jamot, Laure; Weinbach, Jérôme; Chassaing, Nicolas; Team, Technical

doi:10.1111/j.1755-3768.2017.02782

Cited by 2 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, other modeling strategies that integrate graphs (and their topology) into the learning process, such as graph-based CNN [8,16] would need to be explored as well. [8,4] 83.9% Fischer-M, both [8,8,8] 86.8% Fischer-R, centralities [8,8,8,4] 85.8% Fischer-R, modularities [8,8,8,2] 82.1% Fischer-R, both [2,2,2,2] 85.2% Fischer a , centralities [8,2,2] 86.1% Fischer a , modularities [8,2,2] 84.7% Fischer a , both [8,8,4] 84.7% Clinical outcome = 'Disease progression' Fischer-M, centralities [8,8,8,2] 84.3% Fischer-M, modularities [8,8,2] 82.3% Fischer-M, both [4,4,2] 83.7% Fischer-R, centralities [8,8,4] 83.7% Fischer-R, modularities [8,2,2] 79.1% Fischer-R, both [8,8,8,…”

Section: Resultsmentioning

confidence: 99%

“…We can observe a small inverse correlation between the complexity of the architecture and the average performance. Although significant, the average performance of simple models (one hidden layer) is, on average, only marginally better than the average performance of more complex models (at least two hidden layers) (p 1e-7, ∆ bACC ∈ [2,4]).…”

Section: Influence Of the Dnn Architecturementioning

confidence: 98%

“…The cost (linear SVM), gamma (linear and RBF SVM) and number of trees (RF) parameters are optimized using a grid search. The cost and gamma parameters are set to 2 2p , with p ∈ Z, p ∈ [− 4,4]. The number of trees varies between 100 and 10,000.…”

Section: Other Modeling Approachesmentioning

confidence: 99%

“…GEDFN parameter space is explored using a small grid search that always include the default values suggested by the authors. The parameters we optimize are the number of neurons for the second and third layers (∈ {(64, 16), (16,4)}), the learning rate (∈ {1e-4, 1e-2}), the adam optimizer regularization (∈ {T rue, F alse}), the number of epochs (∈ {100, 1000}) and the batch size (∈ {8, 32}). Each GEDFN model is run ten times with different initialization weights and biases.…”

Section: Other Modeling Approachesmentioning

confidence: 99%

See 3 more Smart Citations

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

Tranchevent

Azuaje

Rajapakse

2019

Preprint

View full text Add to dashboard Cite

The availability of high-throughput omics datasets from large patient cohorts has allowed the development of methods that aim at predicting patient clinical outcomes, such as survival and disease recurrence. Such methods are also important to better understand the biological mechanisms underlying disease etiology and development, as well as treatment responses. Recently, different predictive models, relying on distinct algorithms (including Support Vector Machines and Random Forests) have been investigated. In this context, deep learning strategies are of special interest due to their demonstrated superior performance over a wide range of problems and datasets. One of the main challenges of such strategies is the "small n large p" problem. Indeed, omics datasets typically consist of small numbers of samples and large numbers of features relative to typical deep learning datasets. Neural networks usually tackle this problem through feature selection or by including additional constraints during the learning process.We propose to tackle this problem with a novel strategy that relies on a graph-based method for feature extraction, coupled with a deep neural network for clinical outcome prediction. The omics data are first represented as graphs whose nodes represent patients, and edges represent correlations between the patients' omics profiles. Topological features, such as centralities, are then extracted from these graphs for every node. Lastly, these features are used as input to train and test various classifiers.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Influence Of the Dnn Architecturementioning

confidence: 98%

Section: Other Modeling Approachesmentioning

confidence: 99%

Section: Other Modeling Approachesmentioning

confidence: 99%

See 2 more Smart Citations

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

Tranchevent

Azuaje

Rajapakse

2019

Preprint

View full text Add to dashboard Cite

show abstract

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

2019

View full text Add to dashboard Cite

BackgroundThe availability of high-throughput omics datasets from large patient cohorts has allowed the development of methods that aim at predicting patient clinical outcomes, such as survival and disease recurrence. Such methods are also important to better understand the biological mechanisms underlying disease etiology and development, as well as treatment responses. Recently, different predictive models, relying on distinct algorithms (including Support Vector Machines and Random Forests) have been investigated. In this context, deep learning strategies are of special interest due to their demonstrated superior performance over a wide range of problems and datasets. One of the main challenges of such strategies is the “small n large p” problem. Indeed, omics datasets typically consist of small numbers of samples and large numbers of features relative to typical deep learning datasets. Neural networks usually tackle this problem through feature selection or by including additional constraints during the learning process.MethodsWe propose to tackle this problem with a novel strategy that relies on a graph-based method for feature extraction, coupled with a deep neural network for clinical outcome prediction. The omics data are first represented as graphs whose nodes represent patients, and edges represent correlations between the patients’ omics profiles. Topological features, such as centralities, are then extracted from these graphs for every node. Lastly, these features are used as input to train and test various classifiers.ResultsWe apply this strategy to four neuroblastoma datasets and observe that models based on neural networks are more accurate than state of the art models (DNN: 85%-87%, SVM/RF: 75%-82%). We explore how different parameters and configurations are selected in order to overcome the effects of the small data problem as well as the curse of dimensionality.ConclusionsOur results indicate that the deep neural networks capture complex features in the data that help predicting patient clinical outcomes.

show abstract

The RaDiCo AC‐OEIL : a French rare disease cohort dedicated to ocular developmental anomalies in children

Cited by 2 publications

References 0 publications

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

A deep neural network approach to predicting clinical outcomes of neuroblastoma patients

Contact Info

Product

Resources

About