An automated machine learning approach to predict brain age from cortical anatomical measures

Abstract

The use of machine learning (ML) algorithms has significantly increased in neuroscience. However, from the vast extent of possible ML algorithms, which one is the optimal model to predict the target variable? What are the hyperparameters for such a model? Given the plethora of possible answers to these questions, in the last years, automated ML (autoML) has been gaining attention. Here, we apply an autoML library called Tree‐based Pipeline Optimisation Tool (TPOT) which uses a tree‐based representation of ML pipelines and conducts a genetic programming‐based approach to find the model and its hyperparameters that more closely predicts the subject's true age. To explore autoML and evaluate its efficacy within neuroimaging data sets, we chose a problem that has been the focus of previous extensive study: brain age prediction. Without any prior knowledge, TPOT was able to scan through the model space and create pipelines that outperformed the state‐of‐the‐art accuracy for Freesurfer‐based models using only thickness and volume information for anatomical structure. In particular, we compared the performance of TPOT (mean absolute error [MAE]: 4.612 ± .124 years) and a relevance vector regression (MAE 5.474 ± .140 years). TPOT also suggested interesting combinations of models that do not match the current most used models for brain prediction but generalise well to unseen data. AutoML showed promising results as a data‐driven approach to find optimal models for neuroimaging applications.

Brain-Age Prediction Using Shallow Machine Learning: Predictive Analytics Competition 2019

“…To address this issue, in this competition we used: (i) grid search strategy, which repeatedly performed the analysis over a set of pre-defined hyperparameters; (ii) a genetic-based method that was performed by TPOT in order to find the most appropriate model and its hyperparameters (i.e., taking into account both precision and complexity). Similarly, to the results reported by Dafflon et al ( 42 ) and the no free lunch principle ( 43 ), we observed that there was not a single model that always had the best performance when predicting age ( Table 3 ). The different models identified by TPOT for each site probably changed due to biological (i.e., age range, population heterogeneity) and non-biological factors (i.e., field strength and scanner manufacturer).…”

Brain Age Prediction With Morphological Features Using Deep Neural Networks: Results From Predictive Analytic Competition 2019

“…Several solutions have been proposed to overcome these limitations. As an example, ( 48 ) proposed a completely automated pipeline that can find the most appropriate model for the dataset under analysis and provide a complete comparison with the most commonly used models. Different models and their hyperparameters are extensively tested to provide the optimal model for the training dataset.…”

Automated machine learning to predict the co‐occurrence of isocitrate dehydrogenase mutations and O6‐methylguanine‐DNA methyltransferase promoter methylation in patients with gliomas

“…The ability of TPOT is to identify the best pipeline that can be used to fit the statistical properties of the underlying dataset while controlling for overfitting and reliability. 27 Prior studies have compared the models derived by TPOT with a basic random forest (RF) and relevance vector regression (RVR) and shown that the models from TPOT had a significant improvement than RF or RVR. 24,27 Our findings also showed a relatively high accuracy of prediction and suggested that TPOT is feasible and promising for genotype status prediction.…”