2 results
Identifying neuroanatomical signatures of anorexia nervosa: a multivariate machine learning approach
- L. Lavagnino, F. Amianto, B. Mwangi, F. D'Agata, A. Spalatro, G. B. Zunta-Soares, G. Abbate Daga, P. Mortara, S. Fassino, J. C. Soares
-
- Journal:
- Psychological Medicine / Volume 45 / Issue 13 / October 2015
- Published online by Cambridge University Press:
- 20 May 2015, pp. 2805-2812
-
- Article
- Export citation
-
Background
There are currently no neuroanatomical biomarkers of anorexia nervosa (AN) available to make clinical inferences at an individual subject level. We present results of a multivariate machine learning (ML) approach utilizing structural neuroanatomical scan data to differentiate AN patients from matched healthy controls at an individual subject level.
MethodStructural neuroimaging scans were acquired from 15 female patients with AN (age = 20, s.d. = 4 years) and 15 demographically matched female controls (age = 22, s.d. = 3 years). Neuroanatomical volumes were extracted using the FreeSurfer software and input into the Least Absolute Shrinkage and Selection Operator (LASSO) multivariate ML algorithm. LASSO was ‘trained’ to identify ‘novel’ individual subjects as either AN patients or healthy controls. Furthermore, the model estimated the probability that an individual subject belonged to the AN group based on an individual scan.
ResultsThe model correctly predicted 25 out of 30 subjects, translating into 83.3% accuracy (sensitivity 86.7%, specificity 80.0%) (p < 0.001; χ2 test). Six neuroanatomical regions (cerebellum white matter, choroid plexus, putamen, accumbens, the diencephalon and the third ventricle) were found to be relevant in distinguishing individual AN patients from healthy controls. The predicted probabilities showed a linear relationship with drive for thinness clinical scores (r = 0.52, p < 0.005) and with body mass index (BMI) (r = −0.45, p = 0.01).
ConclusionsThe model achieved a good predictive accuracy and drive for thinness showed a strong neuroanatomical signature. These results indicate that neuroimaging scans coupled with ML techniques have the potential to provide information at an individual subject level that might be relevant to clinical outcomes.
Algebraic Methods for Studying Interactions Between Epidemiological Variables
- F. Ricceri, C. Fassino, G. Matullo, M. Roggero, M.-L. Torrente, P. Vineis, L. Terracini
-
- Journal:
- Mathematical Modelling of Natural Phenomena / Volume 7 / Issue 3 / 2012
- Published online by Cambridge University Press:
- 06 June 2012, pp. 227-252
- Print publication:
- 2012
-
- Article
- Export citation
-
Background
Independence models among variables is one of the most relevant topics in epidemiology, particularly in molecular epidemiology for the study of gene-gene and gene-environment interactions. They have been studied using three main kinds of analysis: regression analysis, data mining approaches and Bayesian model selection. Recently, methods of algebraic statistics have been extensively used for applications to biology. In this paper we present a synthetic, but complete description of independence models in algebraic statistics and a new method of analyzing interactions, that is equivalent to the correction by Markov bases of the Fisher’s exact test.
Methods
We identified the suitable algebraic independence model for describing the dependence of two genetic variables from the occurrence of cancer and exploited the theory of toric varieties and Gröbner basis for developing an exact independence test based on the Diaconis-Sturmfels algorithm. We implemented it in a Maple routine and we applied it to the study of gene-gene interaction in Gen-Air, an European case-control study. We computed the p-value for each pair of genetic variables interacting with disease status and we compared our results with the standard asymptotic chi-square test.
Results
We found an association among COMT Val158Met, APE1 Asp148Glu and bladder cancer (p-value: 0.009). We also found the interaction among TP53 Arg72Pro, GSTP1 Ile105Val and lung cancer (p-value: 0.00035). Leukaemia was observed to significantly interact with the pairs ERCC2 Lys751Gln and RAD51 172 G > T (p-value 0.0072), ERCC2 Lys751Gln and LIG4Thr9Ile (p-value: 0.0095) and APE1 Asp148Glu and GSTP1 Ala114Val (p-value: 0.0036).
Conclusion
Taking advantage of results from theoretical and computational algebra, the method we propose was more selective than other methods in detecting new interactions, and nevertheless its results were consistent with previous epidemiological and functional findings. It also helped us in controlling the multiple comparison problem. In the light of our results, we believe that the epidemiologic study of interactions can benefit of algebraic methods based on properties of toric varieties and Gröbner bases.