Early Breast Cancer Prognosis Prediction and Rule Extraction Using a New Constructive Neural Network Algorithm

Abstract

Abstract. Breast cancer relapse prediction is an important step in the complex decision-making process of deciding the type of treatment to be applied to patients after surgery. Some non-linear models, like neural networks, have been successfully applied to this task but they suffer from the problem of extracting the underlying rules, and knowing how the methods operate can help to a better understanding of the cancer relapse problem. A recently introduced constructive algorithm (DASG) that creates compact neural network architectures is applied to a dataset of early breast cancer patients with the aim of testing the predictive ability of the new method. The DASG method works with Boolean input data and for that reason a transformation procedure was applied to the original data. The degradation in the predictive performance due to the transformation of the data is also analyzed using the new method and other standard algorithms.

Semantic Web-Based Interoperability for Intelligent Agents with PSyKE

“…Symbols adopted by SKE algorithms to represent intelligible knowledge are, for instance, lists or trees of rules [14,24,31,32,33] that can be used to make human-understandable predictions as well as to shed a light on the internal behaviour of a BB model. SKE is a precious resource when dealing with critical application fields -e.g., healthcare [6,13,18], financial forecasting [3,4,43], credit card screening [38], but not only [2,21] -, where it is not acceptable to make decisions on the basis of "blind" AI predictions. For example, consider the case of an autonomous vehicle that does not steer when it is about to collide with a pedestrian.…”

Towards a unified model for symbolic knowledge extraction with hypercube-based methods

“…SKE allows data scientists to associate human-comprehensible, post-hoc explanations [18] to the recommendations or decisions computed by the most common prediction-effective -yet, poorly interpretable -algorithms. For instance, SKE is widely adopted for credit-risk evaluation [3,38], med-ical diagnosis [5,13], credit card screening [36], intrusion detection systems [15], keyword extraction [2], and space mission data prediction [33].…”

The PSyKE Technology for Trustworthy Artificial Intelligence

“…Symbols may consist of comprehensible knowledge-e.g., lists or trees of rules that can be exploited to either derive predictions or to better understand the BB behaviour and, as a further step, as knowledge on which to perform any kind of logical reasoning. Currently, SKE techniques have been already applied in a wide variety of areas, ranging from medical diagnosis [10] to finance [1] and astrophysics [22]. Despite the wide adoption of SKE and the existence of different techniques for extracting symbolic knowledge out of a BB, a unified and generalpurpose software technology supporting such methods and their comparison is currently lacking.…”