Explainable Machine Learning Using SHAP - An Application in Alzheimer’s Disease Diagnosis

Aug 3, 2026

Upload by Aurora

Machine learning models are widely used for analyzing large datasets and extracting meaningful patterns, yet their growing complexity can hinder interpretation of decision-making. Shapley Additive Explanations (SHAP) provides a framework for explaining model outputs by leveraging principles from cooperative game theory. The work first evaluates SHAP’s reliability and studies influences such as feature correlation and feature interactions, supported by tests on synthetic data. The method is then applied to Alzheimer’s disease diagnosis using ADNI data and a random forest model built from structural MRI and amyloid-beta PET features.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages68

Reading Time2 h 51 min

AccessFree to download

GRAU DE MATEMÀTIQUESTreball final de grau Explainable Machine Learning Using SHAP: An Application in Alzheimer’s Disease Diagnosis Autor: Marc Ballestero Ribó Directors: Dra. Agnès Pérez-Millan Dra. Petia Radeva Realitzat a: Departament de Matemàtiques i Informàtica Unitat de Biofísica i Bioenginyeria Barcelona, January 17, 2024 Abstract Machine learning models are a powerful and increasingly ubiquitous tool in modern science, due to their ability to analyse massive amounts of data and extract valuable information. Nonetheless, the increasing complexity of such models usually impedes the interpretation of their decision-making processes. Shapley Additive Explanations (SHAP) is a recently developed explainability method that poses as a suitable candidate to solve this problem. Healthcare is a field where explainability emerges as a critical aspect for the deployment of machine learning techniques. Recently, some studies have started exploring the usage of SHAP on machine learning models for the diagnosis of Alzheimer’s disease (AD) . The aim of this work is, firstly, to conduct a theoretical exploration of SHAP to assess its reliability, and then to apply it in a machine learning model for the diagnosis of AD. A theoretical description of SHAP has been done starting from its basis, which relies on results coming from cooperative game theory. A subsequent analysis has been done to study how factors such as feature correlation or feature interaction may affect the method. Numerical tests with synthetic data have been done to illustrate the previous theoretical discussions. Having theoretically presented and discussed the SHAP method, it has been applied in the context of AD diagnosis. Using publicly available data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset, a random forest algorithm has been implemented to classify AD patients and healthy controls, using features derived from structural magnetic resonance imaging and amyloid-beta positron emission tomography. This analysis has served as a proof of context of the clinical applicability of SHAP. Further work should be done in order to refine and fully adapt the method to the context of AD diagnosis. 2020 Mathematics Subject Classification: 68T01, 68T99, 91A12 Resum Els models d’aprenentatge automàtic són una eina poderosa i cada cop més omnipresent en la ciència moderna, a causa de la seva capacitat per analitzar quantitats massives de dades i extreure’n informació valuosa. No obstant això, la complexitat creixent d’aquests models sol dificultar la interpretació dels seus processos interns de presa de decisions. Shapley Additive Explanations (SHAP) és un mètoded’explicabilitat desenvolupat recentment que es presenta com un candidat adequat per resoldre aquest problema. La medicina és un camp on l’explicabilitat sorgeix com un aspecte crític per al desplegament de tècniques d’aprenentatge automàtic. Recentment, alguns estudishan començat a explorar l’ús de SHAP en models d’aprenentatge automàtic per al diagnòstic de la malaltia d’Alzheimer (MA) . L’objectiu d’aquest treball ésfer una exploració teòrica deSHAP per avaluar-ne la fiabilitat i, subsegüentment, aplicar aquest mètode en un model d’aprenentatge automàtic per al diagnòstic de la MA. Primerament, s’ha fet una descripció teòrica de SHAP partint de la seva fonamentació en resultats provinents de la teoria de jocs cooperatius. Posteriorment, s’ha fet una anàlisi per estudiar com el mètode pot veure’s afectat per factors com ara la correlació de característiques i la interacció entre elles. Finalment, s’han fet proves numèriquesamb dades sintètiques per il·lustrar les discussions teòriques anteriors. Un cop fet aquest estudi teòric, s’ha aplicat el mètode SHAP en el context del diagnòstic de la MA. Utilitzant dades provinents de la iniciativa pública Alzheimer’s Disease Neuorimaging Initiative (ADNI), s’ha implementat un algorisme de classificació random forest per dif

FAQs

What problem does SHAP address in machine learning?

It improves interpretability by explaining how model decisions are formed, mitigating the difficulty caused by complex model behavior.

How is SHAP theoretically investigated in this work?

The study derives SHAP from cooperative game theory, then analyzes how feature correlation and feature interactions can affect the explanation method.

How is SHAP applied to Alzheimer’s disease diagnosis here?

The approach uses publicly available ADNI data to train a random forest classifier distinguishing Alzheimer’s patients from healthy controls using features from structural MRI and amyloid-beta PET.

Aug 3, 2026

Upload by Aurora