مقاله انگلیسی رایگان در مورد هوش مصنوعی قابل توصیف برای سرطان پستان – الزویر 2019

1. Introduction

2. Related works

3. Visual interface

4. Automatic algorithms

5. Application to breast cancer machine-learning datasets

6. Application to real data in breast cancer

7. Discussion

Acknowledgements

Appendix A. Supplementary data

References

Abstract

Case-Based Reasoning (CBR) is a form of analogical reasoning in which the solution for a (new) query case is determined using a database of previous known cases with their solutions. Cases similar to the query are retrieved from the database, and then their solutions are adapted to the query. In medicine, a case usually corresponds to a patient and the problem consists of classifying the patient in a class of diagnostic or therapy. Compared to “black box” algorithms such as deep learning, the responses of CBR systems can be justified easily using the similar cases as examples. However, this possibility is often under-exploited and the explanations provided by most CBR systems are limited to the display of the similar cases.

In this paper, we propose a CBR method that can be both executed automatically as an algorithm and presented visually in a user interface for providing visual explanations or for visual reasoning. After retrieving similar cases, a visual interface displays quantitative and qualitative similarities between the query and the similar cases, so as one can easily classify the query through visual reasoning, in a fully explainable manner. It combines a quantitative approach (visualized by a scatter plot based on Multidimensional Scaling in polar coordinates, preserving distances involving the query) and a qualitative approach (set visualization using rainbow boxes). We applied this method to breast cancer management. We showed on three public datasets that our qualitative method has a classification accuracy comparable to k-Nearest Neighbors algorithms, but is better explainable. We also tested the proposed interface during a small user study. Finally, we apply the proposed approach to a real dataset in breast cancer. Medical experts found the visual approach interesting as it explains why cases are similar through the visualization of shared patient characteristics.

Introduction

CBR has been applied to many domains, including medicine [3]. In a medical CBR system, a case usually corresponds to a patient and the problem to solve typically consists of classifying a new patient according to various classes. For a diagnostic system targeting a given disorder, there are commonly two classes: healthy vs diseased. For a therapeutic system, there are several classes corresponding to the various categories of possible treatments. The case database contains previous patients, for which the diagnosis or the treatment is known. Many therapeutic decision support systems implement evidence-based clinical practice guidelines [4]. These systems are therefore knowledgebased rather than case-based. However, CBR is still an interesting approach for patients that are not covered by clinical practice guidelines (they can represent up to 45% of patients [5]), or when guideline’s recommendations cannot be applied, e.g. due to contraindications or when the patient refuses the recommended therapy. Moreover, CBR and guideline-based approaches can also be combined together [6].

Many data-driven classification approaches in artificial intelligence suffer from a lack of explainability; this is particularly true for “black box” approaches like deep learning, e.g. IBM Watson, which has been recently applied to breast cancer therapy [7]. However, in medical systems, black boxes are usually not well-appreciated by physicians since they prefer to understand how the system produces a recommendation [8], and automatic decision-support systems are often perceived as a threat and a loss of control [9]. Indeed, years ago, explainability was ranked by physicians as the most desirable feature of a clinical decision support system [10]. Today, in France, the recent Villani report [11] on artificial intelligence recommends to “open the black-box of artificial intelligence”, with a special focus on medicine.