Several works in the literature use propositional (“black box”) approaches to generate prediction models. In this work we employ the Inductive Logic Programming technique, whose prediction model is based on first order rules, to the domain of breast cancer. These rules have the advantage of being interpretable and convenient to be used as a common language between the computer scientists and the medical experts. We also explore the relevance of some of variables usually collected to predict breast cancer. We compare our results with a propositional classifier that was considered best for the same dataset studied in this paper.

Mammoclass is a web tool that allows users to enter a small set of variable values that describe a finding in a mammography, and produces a probability of this finding being malignant or benign. The tool requires that the user types in every variable a value in order to perform a prediction. In this work, we present a speech-to-text interface integrated to MammoClass that allows radiologists to speak up a mammography report instead of typing it in. This new MammoClass module can take audio content, transcribe it into written words, and automatically extract the variable values by applying a parser to the recognized text. Results of spoken mammography reports show that the same variables are extracted for both types of input: typed in or dictated text.

Predicting breast cancer risk has long been a goal of medical research in the pursuit of precision medicine. The goal of this study is to develop novel penalized methods to improve breast cancer risk prediction by leveraging structure information in electronic health records. We conducted a retrospective case- control study, garnering 49 mammography descriptors and 77 high- frequency/low-penetrance single-nucleotide polymorphisms (SNPs) from an existing personalized medicine data repository. Structured mammography reports and breast imaging features have long been part of a standard electronic health record (EHR), and genetic markers likely will be in the near future. Lasso and its variants are widely used approaches to integrated learning and feature selection, and our methodological contribution is to incorporate the dependence structure among the features into these approaches. More specifically, we propose a new methodology by combining group penalty and $\ell^p$ ($1\leq p\leq2$) fusion penalty to improve breast cancer risk prediction, taking into account structure information in mammography descriptors and SNPs. We demonstrate that our method provides benefits that are both statistically significant and potentially significant to people's lives.

Probabilistic Inductive Logic Programming (PILP) is a relatively unexplored area of Statistical Relational Learning which extends classic Inductive Logic Programming (ILP). Within this scope, we introduce SkILL, a Stochastic Inductive Logic Learner, which takes probabilistic annotated data and produces First Order Logic (FOL) theories. Data in several domains such as medicine and bioinformatics have an inherent degree of uncertainty, and because SkILL can handle this type of data, the models produced for these areas are closer to reality. SkILL can then use probabilistic data to extract non-trivial knowledge from databases, and also address efficiency issues by introducing an efficient search strategy for finding hypotheses in PILP environments. SkILL's capabilities are demonstrated using a real world medical dataset in the breast cancer domain.

Hidden Markov random fields (HMRFs) are conventionally assumed to be homogeneous in the sense that the potential functions are invariant across different sites. However in some biological applications, it is desirable to make HMRFs heterogeneous, especially when there exists some background knowledge about how the potential functions vary. We formally define heterogeneous HMRFs and propose an EM algorithm whose M-step combines a contrastive divergence learner with a kernel smoothing step to incorporate the background knowledge. Simulations show that our algorithm is effective for learning heterogeneous HMRFs and outperforms alternative binning methods. We learn a heterogeneous HMRF in a real-world study.

It has been shown that graphical models can be used to leverage the dependence in large-scale multiple testing problems with significantly improved performance (Sun & Cai, 2009; Liu et al., 2012). These graphical models are fully parametric and require that we know the parameterization of f1, the density function of the test statistic under the alternative hypothesis. However in practice, f1 is often heterogeneous, and cannot be estimated with a simple parametric distribution. We propose a novel semiparametric approach for multiple testing under dependence, which estimates f1 adaptively. This semiparametric approach exactly generalizes the local FDR procedure (Efron et al., 2001) and connects with the BH procedure (Benjamini & Hochberg, 1995). A variety of simulations show that our semiparametric approach outperforms classical procedures which assume independence and the parametric approaches which capture dependence.

Bayesian network structures are usually built using only the data and starting from an empty network or from a naïve Bayes structure. Very often, in some domains, like medicine, a prior structure knowledge is already known. This structure can be automatically or manually refined in search for better performance models. In this work, we take Bayesian networks built by specialists and show that minor perturbations to this original network can yield better classifiers with a very small computational cost, while maintaining most of the intended meaning of the original model.

When mammography reveals a suspicious finding, a core needle biopsy is usually recommended. In 5% to 15% of these cases, the biopsy diagnosis is non-definitive and a more invasive surgical excisional biopsy is recommended to confirm a diagnosis. The majority of these cases will ultimately be proven benign. The use of excisional biopsy for diagnosis negatively impacts patient quality of life and increases costs to the healthcare system. In this work, we employ a multi-relational machine learning approach to predict when a patient with a non-definitive core needle biopsy diagnosis need not undergo an excisional biopsy procedure because the risk of malignancy is low.

In this work we build the first BI-RADS parser for Portuguese free texts, modeled after existing approaches to extract BI-RADS features from English medical records. Our concept finder uses a semantic grammar based on the BIRADS lexicon and on iterative transferred expert knowledge. We compare the performance of our algorithm to manual annotation by a specialist in mammography. Our results show that our parser's performance is comparable to the manual method.

Because breast tissue composition partially predicts breast cancer risk, classification of mammography reports by breast tissue composition is important from both a scientific and clinical perspective. A method is presented for using the unstructured text of mammography reports to classify them into BI-RADS breast tissue composition categories. An algorithm that uses regular expressions to automatically determine BI-RADS breast tissue composition classes for unstructured mammography reports was developed. The algorithm assigns each report to a single BI-RADS composition class: 'fatty', 'fibroglandular', 'heterogeneously dense', 'dense', or 'unspecified'. We evaluated its performance on mammography reports from two different institutions. The method achieves >99% classification accuracy on a test set of reports from the Marshfield Clinic (Wisconsin) and Stanford University. Since large-scale studies of breast cancer rely heavily on breast tissue composition information, this method could facilitate this research by helping mine large datasets to correlate breast composition with other covariates.

Breast screening is the regular examination of a woman's breasts to find breast cancer earlier. The sole exam approved for this purpose is mammography. Usually, findings are annotated through the Breast Imaging Reporting and Data System (BIRADS) created by the American College of Radiology. The BIRADS system determines a standard lexicon to be used by radiologists when studying each finding. Although the lexicon is standard, the annotation accuracy of the findings depends on the experience of the radiologist. Moreover, the accuracy of the classification of a mammography is also highly dependent on the expertise of the radiologist. A correct classification is paramount due to economical and humanitarian reasons. The main goal of this work is to produce machine learning models that predict the outcome of a mammography from a reduced set of annotated mammography findings. In the study we used a data set consisting of 348 consecutive breast masses that underwent image guided or surgical biopsy performed between October 2005 and December 2007 on 328 female subjects. The main conclusions are threefold: (1) automatic classification of a mammography, independent on information about mass density, can reach equal or better results than the classification performed by a physician; (2) mass density seems to be a good indicator of malignancy, as previous studies suggested; (3) a machine learning model can predict mass density with a quality as good as the specialist blind to biopsy, which is one of our main contributions. Our model can predict malignancy in the absence of the mass density attribute, since we can fill up this attribute using our mass density predictor.