Abstract. Conformal predictors represent a new flexible framework that outputs region predictions with a guaranteed error rate. Efficiency of such predictions depends on the nonconformity measure that underlies the predictor. In this work we designed new nonconformity measures based on a random forest classifier. Experiments demonstrate that proposed conformal predictors are more efficient than current benchmarks on noisy mass spectrometry data (and at least as efficient on other type of data) while maintaining the property of validity: they output fewer multiple predictions, and the ratio of mistakes does not exceed the preset level. When forced to produce singleton predictions, the designed conformal predictors are at least as accurate as the benchmarks and sometimes significantly outperform them.

Background: The serum peptidome may be a valuable source of diagnostic cancer biomarkers. Previous mass spectrometry (MS) studies have suggested that groups of related peptides discriminatory for different cancer types are generated ex vivo from abundant serum proteins by tumor-specific exopeptidases. We tested 2 complementary serum profiling strategies to see if similar peptides could be found that discriminate ovarian cancer from benign cases and healthy controls. Methods: We subjected identically collected and processed serum samples from healthy volunteers and patients to automated polypeptide extraction on octadecylsilane-coated magnetic beads and separately on ZipTips before MALDI-TOF MS profiling at 2 centers. The 2 platforms were compared and case control profiling data analyzed to find altered MS peak intensities. We tested models built from training datasets for both methods for their ability to classify a blinded test set. Results: Both profiling platforms had CVs of approximately 15% and could be applied for high-throughput analysis of clinical samples. The 2 methods generated overlapping peptide profiles, with some differences in peak intensity in different mass regions. In cross-validation, models from training data gave diagnostic accuracies up to 87% for discriminating malignant ovarian cancer from healthy controls and up to 81% for discriminating malignant from benign samples. Diagnostic accuracies up to 71% (malignant vs healthy) and up to 65% (malignant vs benign) were obtained when the models were validated on the blinded test set. Conclusions: For ovarian cancer, altered MALDI-TOF MS peptide profiles alone cannot be used for accurate diagnoses.

This paper describes the methodology of providing multiprobability predictions for proteomic mass spectrometry data. The methodology is based on a newly developed machine learning framework called Venn machines. Is allows to output a valid probability interval. The methodology is designed for mass spectrometry data. For demonstrative purposes, we applied this methodology to MALDI-TOF data sets in order to predict the diagnosis of heart disease and early diagnoses of ovarian cancer and breast cancer. The experiments showed that probability intervals are narrow, that is, the output of the multiprobability predictor is similar to a single probability distribution. In addition, probability intervals produced for heart disease and ovarian cancer data were more accurate than the output of corresponding probability predictor. When Venn machines were forced to make point predictions, the accuracy of such predictions is for the most data better than the accuracy of the underlying algorithm that outputs single probability distribution of a label. Application of this methodology to MALDI-TOF data sets empirically demonstrates the validity. The accuracy of the proposed method on ovarian cancer data rises from 66.7%11 months in advance of the moment of diagnosis to up to 90.2 % at the moment of diagnosis. The same approach has been applied to heart disease data without time dependency, although the achieved accuracy was not as high (up to 69.9 %). The methodology allowed us to confirm mass spectrometry peaks previously identified as carrying statistically significant information for discrimination between controls and cases.

We construct prediction intervals for the linear regression model with IID errors with a known distribution, not necessarily Gaussian. The coverage probability of our prediction intervals is equal to the nominal confidence level not only unconditionally but also conditionally given a natural σ-algebra of invariant events. This implies, in particular, the perfect calibration of our prediction intervals in the online mode of prediction.

The work describes an application of a recently developed machine-learning technique called Mondrian predictors to risk assessment of ovarian and breast cancers. The analysis is based on mass spectrometry profiling of human serum samples that were collected in the United Kingdom Collaborative Trial of Ovarian Cancer Screening. The work describes the technique and presents the results of classification (diagnosis) and the corresponding measures of confidence of the diagnostics. The main advantage of this approach is a proven validity of prediction. The work also describes an approach to improve early diagnosis of ovarian and breast cancers since the data in the United Kingdom Collaborative Trial of Ovarian Cancer Screening were collected over a period of 7 years and do allow to make observations of changes in human serum over that period of time. Significance of improvement is confirmed statistically (for up to 11 months for ovarian cancer and 9 months for breast cancer). In addition, the methodology allowed us to pinpoint