Are Machine Learning Algorithms More Accurate in Predicting Vegetable and Fruit Consumption Than Traditional Statistical Models? An Exploratory Analysis.

Machine learning (ML) algorithms may help better understand the complex interactions among factors that influence dietary choices and behaviors. The aim of this study was to explore whether ML algorithms are more accurate than traditional statistical models in predicting vegetable and fruit (VF) consumption. A large array of features (2,452 features from 525 variables) encompassing individual and environmental information related to dietary habits and food choices in a sample of 1,147 French-speaking adult men and women was used for the purpose of this study. Adequate VF consumption, which was defined as 5 servings/d or more, was measured by averaging data from three web-based 24 h recalls and used as the outcome to predict. Nine classification ML algorithms were compared to two traditional statistical predictive models, logistic regression and penalized regression (Lasso). The performance of the predictive ML algorithms was tested after the implementation of adjustments, including normalizing the data, as well as in a series of sensitivity analyses such as using VF consumption obtained from a web-based food frequency questionnaire (wFFQ) and applying a feature selection algorithm in an attempt to reduce overfitting. Logistic regression and Lasso predicted adequate VF consumption with an accuracy of 0.64 (95% confidence interval [CI]: 0.58-0.70) and 0.64 (95%CI: 0.60-0.68) respectively. Among the ML algorithms tested, the most accurate algorithms to predict adequate VF consumption were the support vector machine (SVM) with either a radial basis kernel or a sigmoid kernel, both with an accuracy of 0.65 (95%CI: 0.59-0.71). The least accurate ML algorithm was the SVM with a linear kernel with an accuracy of 0.55 (95%CI: 0.49-0.61). Using dietary intake data from the wFFQ and applying a feature selection algorithm had little to no impact on the performance of the algorithms. In summary, ML algorithms and traditional statistical models predicted adequate VF consumption with similar accuracies among adults. These results suggest that additional research is needed to explore further the true potential of ML in predicting dietary behaviours that are determined by complex interactions among several individual, social and environmental factors.

On the robustness of generalization of drug-drug interaction models.

BACKGROUND: Deep learning methods are a proven commodity in many fields and endeavors. One of these endeavors is predicting the presence of adverse drug-drug interactions (DDIs). The models generated can predict, with reasonable accuracy, the phenotypes arising from the drug interactions using their molecular structures. Nevertheless, this task requires improvement to be truly useful. Given the complexity of the predictive task, an extensive benchmarking on structure-based models for DDIs prediction was performed to evaluate their drawbacks and advantages. RESULTS: We rigorously tested various structure-based models that predict drug interactions using different splitting strategies to simulate different real-world scenarios. In addition to the effects of different training and testing setups on the robustness and generalizability of the models, we then explore the contribution of traditional approaches such as multitask learning and data augmentation. CONCLUSION: Structure-based models tend to generalize poorly to unseen drugs despite their ability to identify new DDIs among drugs seen during training accurately. Indeed, they efficiently propagate information between known drugs and could be valuable for discovering new DDIs in a database. However, these models will most probably fail when exposed to unknown drugs. While multitask learning does not help in our case to solve the problem, the use of data augmentation does at least mitigate it. Therefore, researchers must be cautious of the bias of the random evaluation scheme, especially if their goal is to discover new DDIs.

Machine learning analysis identifies genes differentiating triple negative breast cancers.

Triple negative breast cancer (TNBC) is one of the most aggressive form of breast cancer (BC) with the highest mortality due to high rate of relapse, resistance, and lack of an effective treatment. Various molecular approaches have been used to target TNBC but with little success. Here, using machine learning algorithms, we analyzed the available BC data from the Cancer Genome Atlas Network (TCGA) and have identified two potential genes, TBC1D9 (TBC1 domain family member 9) and MFGE8 (Milk Fat Globule-EGF Factor 8 Protein), that could successfully differentiate TNBC from non-TNBC, irrespective of their heterogeneity. TBC1D9 is under-expressed in TNBC as compared to non-TNBC patients, while MFGE8 is over-expressed. Overexpression of TBC1D9 has a better prognosis whereas overexpression of MFGE8 correlates with a poor prognosis. Protein-protein interaction analysis by affinity purification mass spectrometry (AP-MS) and proximity biotinylation (BioID) experiments identified a role for TBC1D9 in maintaining cellular integrity, whereas MFGE8 would be involved in various tumor survival processes. These promising genes could serve as biomarkers for TNBC and deserve further investigation as they have the potential to be developed as therapeutic targets for TNBC.