Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease

Karim Haddada; Mohamed Ibn Khedher; Olfa Jemai; Sarra Iben Khedher; Mounim A. El-Yacoubi

doi:10.1109/HSI61632.2024.10613551

Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease

Karim Haddada
, Mohamed Ibn Khedher
, Olfa Jemai
, Sarra Iben Khedher
, Mounim A. El-Yacoubi

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Alzheimer's disease (AD) is a chronic and irreversible neurological disorder, making early detection essential for managing its progression. This study investigates the coherence of SHAP values with medical scientific truth. It examines three types of features: clinical, demographic, and FreeSurfer extracted from MRI scans. A set of six ML classifiers are investigated for their interpretability levels. This study is validated on the OASIS-3 dataset with binary classification. The results show that clinical data outperforms the others, with a margin of 14% over FreeSurfer features, the second-best features. In the case of clinical features, the explanations provided by the tree-based classifiers consistently align with medical insights. This comparison was calculated using the Kendall Tau distance.

Original language	English
Title of host publication	2024 16th International Conference on Human System Interaction, HSI 2024
Publisher	IEEE Computer Society
ISBN (Electronic)	9798350362916
DOIs	https://doi.org/10.1109/HSI61632.2024.10613551
Publication status	Published - 1 Jan 2024
Event	16th International Conference on Human System Interaction, HSI 2024 - Paris, France Duration: 8 Jul 2024 → 11 Jul 2024

Publication series

Name	International Conference on Human System Interaction, HSI
ISSN (Print)	2158-2246
ISSN (Electronic)	2158-2254

Conference

Conference	16th International Conference on Human System Interaction, HSI 2024
Country/Territory	France
City	Paris
Period	8/07/24 → 11/07/24

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Early Alzheimer Disease
Explainable AI
Interpretability
Machine Learning
SHAP

Access to Document

10.1109/HSI61632.2024.10613551

Cite this

Haddada, K., Khedher, M. I., Jemai, O., Khedher, S. I., & El-Yacoubi, M. A. (2024). Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease. In 2024 16th International Conference on Human System Interaction, HSI 2024 (International Conference on Human System Interaction, HSI). IEEE Computer Society. https://doi.org/10.1109/HSI61632.2024.10613551

@inproceedings{bb806fd8fe91483a9ec42e4edf87aa1e,

title = "Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease",

abstract = "Alzheimer's disease (AD) is a chronic and irreversible neurological disorder, making early detection essential for managing its progression. This study investigates the coherence of SHAP values with medical scientific truth. It examines three types of features: clinical, demographic, and FreeSurfer extracted from MRI scans. A set of six ML classifiers are investigated for their interpretability levels. This study is validated on the OASIS-3 dataset with binary classification. The results show that clinical data outperforms the others, with a margin of 14\% over FreeSurfer features, the second-best features. In the case of clinical features, the explanations provided by the tree-based classifiers consistently align with medical insights. This comparison was calculated using the Kendall Tau distance.",

keywords = "Early Alzheimer Disease, Explainable AI, Interpretability, Machine Learning, SHAP",

author = "Karim Haddada and Khedher, \{Mohamed Ibn\} and Olfa Jemai and Khedher, \{Sarra Iben\} and El-Yacoubi, \{Mounim A.\}",

note = "Publisher Copyright: {\textcopyright} 2024 IEEE.; 16th International Conference on Human System Interaction, HSI 2024 ; Conference date: 08-07-2024 Through 11-07-2024",

year = "2024",

month = jan,

day = "1",

doi = "10.1109/HSI61632.2024.10613551",

language = "English",

series = "International Conference on Human System Interaction, HSI",

publisher = "IEEE Computer Society",

booktitle = "2024 16th International Conference on Human System Interaction, HSI 2024",

}

Haddada, K, Khedher, MI, Jemai, O, Khedher, SI & El-Yacoubi, MA 2024, Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease. in 2024 16th International Conference on Human System Interaction, HSI 2024. International Conference on Human System Interaction, HSI, IEEE Computer Society, 16th International Conference on Human System Interaction, HSI 2024, Paris, France, 8/07/24. https://doi.org/10.1109/HSI61632.2024.10613551

Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease. / Haddada, Karim; Khedher, Mohamed Ibn; Jemai, Olfa et al.
2024 16th International Conference on Human System Interaction, HSI 2024. IEEE Computer Society, 2024. (International Conference on Human System Interaction, HSI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease

AU - Haddada, Karim

AU - Khedher, Mohamed Ibn

AU - Jemai, Olfa

AU - Khedher, Sarra Iben

AU - El-Yacoubi, Mounim A.

PY - 2024/1/1

Y1 - 2024/1/1

N2 - Alzheimer's disease (AD) is a chronic and irreversible neurological disorder, making early detection essential for managing its progression. This study investigates the coherence of SHAP values with medical scientific truth. It examines three types of features: clinical, demographic, and FreeSurfer extracted from MRI scans. A set of six ML classifiers are investigated for their interpretability levels. This study is validated on the OASIS-3 dataset with binary classification. The results show that clinical data outperforms the others, with a margin of 14% over FreeSurfer features, the second-best features. In the case of clinical features, the explanations provided by the tree-based classifiers consistently align with medical insights. This comparison was calculated using the Kendall Tau distance.

AB - Alzheimer's disease (AD) is a chronic and irreversible neurological disorder, making early detection essential for managing its progression. This study investigates the coherence of SHAP values with medical scientific truth. It examines three types of features: clinical, demographic, and FreeSurfer extracted from MRI scans. A set of six ML classifiers are investigated for their interpretability levels. This study is validated on the OASIS-3 dataset with binary classification. The results show that clinical data outperforms the others, with a margin of 14% over FreeSurfer features, the second-best features. In the case of clinical features, the explanations provided by the tree-based classifiers consistently align with medical insights. This comparison was calculated using the Kendall Tau distance.

KW - Early Alzheimer Disease

KW - Explainable AI

KW - Interpretability

KW - Machine Learning

KW - SHAP

U2 - 10.1109/HSI61632.2024.10613551

DO - 10.1109/HSI61632.2024.10613551

M3 - Conference contribution

AN - SCOPUS:85201568628

T3 - International Conference on Human System Interaction, HSI

BT - 2024 16th International Conference on Human System Interaction, HSI 2024

PB - IEEE Computer Society

T2 - 16th International Conference on Human System Interaction, HSI 2024

Y2 - 8 July 2024 through 11 July 2024

ER -

Assessing the Interpretability of Machine Learning Models in Early Detection of Alzheimer's Disease

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Fingerprint

Cite this