See the published version of this preprint at BMC Medical Informatics and Decision Making.
This is a preprint, a preliminary version of a manuscript that has not completed peer review at a journal. Research Square does not conduct peer review prior to posting preprints. The posting of a preprint on this server should not be interpreted as an endorsement of its validity or suitability for dissemination as established information or for guiding clinical practice.
Learn more about In Review
Technical advance
Julian Hatwell
Birmingham City University
Corresponding Author
ORCiD: https://orcid.org/0000-0001-6589-5165
License:
This work is licensed under a CC BY 4.0 License. Read Full License
Background
Computer Aided Diagnostics (CAD) can support medical practitioners to make critical decisions about their patients' disease conditions. Practitioners require access to the chain of reasoning behind CAD to build trust in the CAD advice and to supplement their own expertise. Yet, CAD systems might be based on black box machine learning (ML) models and high dimensional data sources (electronic health records, MRI scans, cardiotocograms, etc). These foundations make interpretation and explanation of the CAD advice very challenging. This challenge is recognised throughout the machine learning research community. eXplainable Artificial Intelligence (XAI) is emerging as one of the most important research areas of recent years because it addresses the interpretability and trust concerns of critical decision makers, including those in clinical and medical practice. Methods
In this work, we focus on AdaBoost, a black box ML model that has been widely adopted in the CAD literature. We address the challenge -- to explain AdaBoost classification -- with a novel algorithm that extracts simple, logical rules from AdaBoost models. Our algorithm, Adaptive-Weighted High Importance Path Snippets (Ada-WHIPS), makes use of AdaBoost's adaptive classifier weights. Using a novel formulation, Ada-WHIPS uniquely redistributes the weights among individual decision nodes of the internal decision trees (DT) of the AdaBoost model. Then, a simple heuristic search of the weighted nodes finds a single rule that dominated the model's decision. We compare the explanations generated by our novel approach with the state of the art in an experimental study. We evaluate the derived explanations with simple statistical tests of well-known quality measures, precision and coverage, and a novel measure stability that is better suited to the XAI setting .
Keywords
Explainable AI; Computer Aided Diagnostics; AdaBoost; Black Box Problem; Interpretability
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13
This preprint is available for download as a PDF.
Loading...
Badges
Prescreen
This manuscript passed our Prescreen assessment
Complete author information
Appropriate declaration statements
Potential risks to human health
Prescreen
Peer Review Timeline
CURRENT STATUS: Published
View the published article at BMC Medical Informatics and Decision Making.
No community comments so far
Submission checks complete
On 27 Sep, 2020
Editorial decision: Accept
On 22 Jul, 2020
No comments provided
Editorial decision: Minor revision
On 21 Jul, 2020
Version 3
Posted 24 Sep, 2020
No comments provided
Editorial decision: Minor revision
On 14 Jul, 2020
Editor assigned
On 17 Jun, 2020
Submission checks complete
On 16 Jun, 2020
Editor invited
On 16 Jun, 2020
No comments provided
Editorial decision: Minor revision
On 04 Jun, 2020
Review #5 received
Received 22 May, 2020
Review #3 received
Received 22 May, 2020
Reviewer #5 agreed
On 18 May, 2020
Reviewer #4 agreed
On 17 May, 2020
Reviewer #3 agreed
On 21 Apr, 2020
Review #2 received
Received 24 Mar, 2020
Review #1 received
Received 24 Mar, 2020
Reviewer #2 agreed
On 28 Feb, 2020
Reviewer #1 agreed
On 26 Feb, 2020
Reviewers invited
Invitations sent on 11 Feb, 2020
Editor assigned
On 10 Feb, 2020
Editor invited
On 09 Feb, 2020
Submission checks complete
On 06 Feb, 2020
No comments provided
Review #2 received
Received 18 Jan, 2020
Editorial decision: Major revision
On 18 Jan, 2020
Review #1 received
Received 10 Jan, 2020
Reviewer #3 agreed
On 09 Jan, 2020
Reviewer #2 agreed
On 06 Jan, 2020
Reviewer #1 agreed
On 31 Dec, 2019
Reviewers invited
Invitations sent on 31 Dec, 2019
Editor assigned
On 18 Dec, 2019
Submission checks complete
On 17 Dec, 2019
Editor invited
On 16 Dec, 2019
First submitted
On 13 Dec, 2019
Metrics
Comments: 0
PDF Downloads: ...
HTML Views: ...
Subject Areas
Medical Informatics
Learn more about our company.
See the published version of this preprint at BMC Medical Informatics and Decision Making.
This is a preprint, a preliminary version of a manuscript that has not completed peer review at a journal. Research Square does not conduct peer review prior to posting preprints. The posting of a preprint on this server should not be interpreted as an endorsement of its validity or suitability for dissemination as established information or for guiding clinical practice.
Learn more about In Review
Technical advance
Julian Hatwell
Birmingham City University
Corresponding Author
ORCiD: https://orcid.org/0000-0001-6589-5165
Badges
Prescreen
This manuscript passed our Prescreen assessment
Complete author information
Appropriate declaration statements
Potential risks to human health
Prescreen
Peer Review Timeline
CURRENT STATUS: Published
View the published article at BMC Medical Informatics and Decision Making.
No community comments so far
Submission checks complete
On 27 Sep, 2020
Editorial decision: Accept
On 22 Jul, 2020
No comments provided
Editorial decision: Minor revision
On 21 Jul, 2020
Version 3
Posted 24 Sep, 2020
No comments provided
Editorial decision: Minor revision
On 14 Jul, 2020
Editor assigned
On 17 Jun, 2020
Submission checks complete
On 16 Jun, 2020
Editor invited
On 16 Jun, 2020
No comments provided
Editorial decision: Minor revision
On 04 Jun, 2020
Review #5 received
Received 22 May, 2020
Review #3 received
Received 22 May, 2020
Reviewer #5 agreed
On 18 May, 2020
Reviewer #4 agreed
On 17 May, 2020
Reviewer #3 agreed
On 21 Apr, 2020
Review #2 received
Received 24 Mar, 2020
Review #1 received
Received 24 Mar, 2020
Reviewer #2 agreed
On 28 Feb, 2020
Reviewer #1 agreed
On 26 Feb, 2020
Reviewers invited
Invitations sent on 11 Feb, 2020
Editor assigned
On 10 Feb, 2020
Editor invited
On 09 Feb, 2020
Submission checks complete
On 06 Feb, 2020
No comments provided
Review #2 received
Received 18 Jan, 2020
Editorial decision: Major revision
On 18 Jan, 2020
Review #1 received
Received 10 Jan, 2020
Reviewer #3 agreed
On 09 Jan, 2020
Reviewer #2 agreed
On 06 Jan, 2020
Reviewer #1 agreed
On 31 Dec, 2019
Reviewers invited
Invitations sent on 31 Dec, 2019
Editor assigned
On 18 Dec, 2019
Submission checks complete
On 17 Dec, 2019
Editor invited
On 16 Dec, 2019
First submitted
On 13 Dec, 2019
Metrics
Comments: 0
PDF Downloads: ...
HTML Views: ...
Subject Areas
Medical Informatics
License:
This work is licensed under a CC BY 4.0 License. Read Full License
View the published article at BMC Medical Informatics and Decision Making.
Background
Computer Aided Diagnostics (CAD) can support medical practitioners to make critical decisions about their patients' disease conditions. Practitioners require access to the chain of reasoning behind CAD to build trust in the CAD advice and to supplement their own expertise. Yet, CAD systems might be based on black box machine learning (ML) models and high dimensional data sources (electronic health records, MRI scans, cardiotocograms, etc). These foundations make interpretation and explanation of the CAD advice very challenging. This challenge is recognised throughout the machine learning research community. eXplainable Artificial Intelligence (XAI) is emerging as one of the most important research areas of recent years because it addresses the interpretability and trust concerns of critical decision makers, including those in clinical and medical practice. Methods
In this work, we focus on AdaBoost, a black box ML model that has been widely adopted in the CAD literature. We address the challenge -- to explain AdaBoost classification -- with a novel algorithm that extracts simple, logical rules from AdaBoost models. Our algorithm, Adaptive-Weighted High Importance Path Snippets (Ada-WHIPS), makes use of AdaBoost's adaptive classifier weights. Using a novel formulation, Ada-WHIPS uniquely redistributes the weights among individual decision nodes of the internal decision trees (DT) of the AdaBoost model. Then, a simple heuristic search of the weighted nodes finds a single rule that dominated the model's decision. We compare the explanations generated by our novel approach with the state of the art in an experimental study. We evaluate the derived explanations with simple statistical tests of well-known quality measures, precision and coverage, and a novel measure stability that is better suited to the XAI setting .
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13
This preprint is available for download as a PDF.
Loading...