Multi-perspective machine learning MPML - A high-performance and interpretable ensemble method for heart disease prediction

Aug 3, 2026

Upload by Aurelia

Machine Learning (ML) demonstrates strong predictive capability in healthcare, often exceeding human performance in pattern recognition and decision-making. Many high-performing models remain difficult to interpret, creating barriers for clinical trust, accountability, and ethical deployment. The Multi-Perspective machine learning framework (MPML) combines established base classifiers with a structured perspective-based design and an interpretability pipeline. Features are organized into meaningful subsets (perspectives) for both global and instance-level explanations. Applied to a heart disease dataset, MPML improves predictive accuracy while providing accessible explanations for individual patient outcomes, unlike standard ensembles such as Bagging, Boosting, and Random Forest.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages20

Reading Time50 min

AccessFree to download

Introduction

Multi-view and interpretability foundations

Architecture and explanation pipeline

Machine Learning with Applications 23 (2026) 100836 Contents lists available at ScienceDirect Machine Learning with Applications journal [homepage: www.elsevier.com/locate/mlwa](homepage: www.elsevier.com/locate/mlwa) | Multi-perspective machine learning MPML: A high-performance and interpretable ensemble method for heart disease prediction
Sean T Miller a,* , Keaton A Logan b, Ricardo Anderson a, Patricia E Cowell c, Curtis Busby-Earlea, Lisa-Dionne Morris d
a The University of the West Indies Mona, Kingston, Jamaica b Caribbean Maritime University, Kingston, Jamaica
c The University of Sheffield, Sheffield, United Kingdom d The University of Leeds, Leeds, United Kingdom | | | | --- | --- | --- | | A R T I C L E I N F O | | A B S T R A C T | | Keywords: Machine learning Healthcare Explainable AI Predictions
Algorithmic accountability | | Machine Learning (ML) has demonstrated strong predictive capabilities in healthcare, often surpassing human performance in pattern recognition and decision-making. However, many high-performing models lack interpretability, which is critical in clinical settings where understanding and trusting predictions is essential. To achieve our objective, we proposed a Multi-Perspective machine learning framework (MPML) that combines established base classifiers with structured perspective-based design and interpretability pipeline. MPML organises features into meaningful subsets, or perspectives, enabling both global and instance-level interpretability. Unlike traditional ensemble methods such as Bagging, Boosting, and Random Forest, MPML delivers significantly higher-quality predictions across all evaluation metrics while maintaining a transparent structure. Applied to a heart disease dataset, MPML not only improves predictive accuracy but also provides detailed, accessible explanations for individual patient outcomes, advancing the potential for practical and ethical deployment of ML in healthcare. | 1. Introduction Machine Learning (ML) has become a powerful tool in data-driven domains such as healthcare, where accurate predictions and informed decision-making are critical. However, many high-performing ML models function as “black boxes,” offering little transparency into how predictions are made (Rudin, 2019). This lack of interpretability poses significant challenges in domains where trust, accountability, and ethical considerations are paramount. To address this gap, we propose Multi-Perspective Machine Learning (MPML). This ensemble approach integrates multiple established techniques to achieve both high predictive performance and model interpretability. Recent efforts to enhance machine learning in healthcare have increasingly focused on balancing predictive performance with interpretability, a challenge that traditional ensemble methods often fail to address. For example, one author (Topuz et al., 2025) emphasized the gap between highly accurate but opaque ensemble models and the need for interpretable AI in critical healthcare tasks. To address this, various hybrid frameworks have been proposed. Another study (Al-bakri et al., 2025) introduced a meta-learning-based ensemble for Alzheimer’s diagnosis, combining predictive strength with transparent decision pathways. Work done in another study (Awe et al., 2025) demonstrated the use of LIME within ensemble models for malaria diagnosis, enhancing clinician trust in model outputs. Similarly, another group of researchers (Acharya et al., 2025) developed a stacking-based XAI approach for diabetes classification, improving interpretability without sacrificing accuracy. In contrast to these approaches, MPML provides a principled integration of multiple perspectives (feature groups formed from statistical correlations and expert knowledge) yielding not only higher predictive power but also inherently interpretable model behaviour. This allows domain experts to trace predictions back to relevant features and perspective

FAQs

Why is interpretability important for healthcare ML models?

High-performing clinical models often act as “black boxes,” which limits transparency. Interpretability supports trust, accountability, and ethical decision-making in clinical settings.

What is the core idea behind MPML?

MPML is an ensemble framework that organizes features into meaningful subsets called perspectives. This structure supports both high predictive performance and interpretability.

How does MPML provide explanations for heart disease predictions?

MPML isolates feature groups and their individual contributions, enabling global and instance-level interpretability. The design adapts concepts inspired by LIME and SHAP to generate accessible explanations for patient outcomes.

Aug 3, 2026

Upload by Aurelia