On the use of double cross-validation for the combination of proteomic mass spectral data for enhanced diagnosis and prediction

B.J.A. Mertens (1) , Y.E.M. van Der Burgt , B. Velstra , W.E. Mesker , R.A.E.M. Tollenaar , A.M. Deelder

1 Departments of Medical Statistics

We consider a proteomic mass spectrometry case-control study for the calibration of a diagnostic rule for the detection of early-stage breast cancer. For each patient, a pair of two distinct mass spectra is recorded, each of which derived from a different prior fractionation procedure on the available patient serum. We propose a procedure to combine the distinct spectral expressions from patients for the calibration of a diagnostic discriminant rule. This is achieved by first calibrating two distinct prediction rules separately, each of which on only one of the two available spectral data sources. A double cross-validatory approach is used to summarize the available spectral data using the two classifiers to posterior class probabilities, on which a combined predictor can be calibrated.

Mots clés: Clinical proteomics; Mass spectrometry; Predictive data fusion; Diagnosis; Double cross-validation; Classification; Model combination

Domaines: Bio-informatique [q-bio.QM]; Bio-Informatique, Biologie Systémique [q-bio.QM]

B.J.A. Mertens, Y.E.M. van Der Burgt, B. Velstra, W.E. Mesker, R.A.E.M. Tollenaar, et al.. On the use of double cross-validation for the combination of proteomic mass spectral data for enhanced diagnosis and prediction. Statistics and Probability Letters, 2011, 81 (7), pp.759. ⟨10.1016/j.spl.2011.02.037⟩. ⟨hal-00746098⟩