Article
An Adaptive Deep Ensemble Learning Method for Dynamic
Evolving Diagnostic Task Scenarios
Kaixiang Su
1
, Jiao Wu
2
, Dongxiao Gu
1,3,
*, Shanlin Yang
1,3
, Shuyuan Deng
4
and Aida K. Khakimova
5
Citation: Su, K.; Wu, J.; Gu, D.; Yang,
S.; Deng, S.; Khakimova, A.K. An
Adaptive Deep Ensemble Learning
Method for Dynamic Evolving
Diagnostic Task Scenarios. Diagnostics
2021, 11, 2288. https://doi.org/
10.3390/diagnostics11122288
Academic Editor: Keun Ho Ryu
Received: 19 November 2021
Accepted: 6 December 2021
Published: 7 December 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
School of Management, Hefei University of Technology, Hefei 230009, China;
2018110745@mail.hfut.edu.cn (K.S.); yangsl@hfut.edu.cn (S.Y.)
2
School of Business, Northern Illinois University, DeKalb, IL 60115, USA; jwu3@niu.edu
3
Key Laboratory of Process Optimization and Intelligent Decision-Making of Ministry of Education,
Hefei 230009, China
4
SpotHero, Chicago, IL 60603, USA; lance0108@gmail.com
5
Scientific-Research Center for Physical-Technical Informatics, Russian New University, Radio St., 22,
105005 Moscow, Russia; aida_khatif@mail.ru
* Correspondence: gudongxiao@hfut.edu.cn
Abstract:
Increasingly, machine learning methods have been applied to aid in diagnosis with good
results. However, some complex models can confuse physicians because they are difficult to under-
stand, while data differences across diagnostic tasks and institutions can cause model performance
fluctuations. To address this challenge, we combined the Deep Ensemble Model (DEM) and tree-
structured Parzen Estimator (TPE) and proposed an adaptive deep ensemble learning method
(TPE-DEM) for dynamic evolving diagnostic task scenarios. Different from previous research that
focuses on achieving better performance with a fixed structure model, our proposed model uses TPE
to efficiently aggregate simple models more easily understood by physicians and require less training
data. In addition, our proposed model can choose the optimal number of layers for the model and
the type and number of basic learners to achieve the best performance in different diagnostic task
scenarios based on the data distribution and characteristics of the current diagnostic task. We tested
our model on one dataset constructed with a partner hospital and five UCI public datasets with
different characteristics and volumes based on various diagnostic tasks. Our performance evaluation
results show that our proposed model outperforms other baseline models on different datasets.
Our study provides a novel approach for simple and understandable machine learning models in
tasks with variable datasets and feature sets, and the findings have important implications for the
application of machine learning models in computer-aided diagnosis.
Keywords:
adaptive deep ensemble learning; dynamic evolving diagnosis; intelligent health knowl-
edge discovery; personalized health management
1. Introduction
Many different factors are often taken into account when diagnosing a disease. The
complexity of the disease (such as the risk levels associated with multiple diseases) and the
diagnostic knowledge available to the physician [
1
,
2
] can influence the correct diagnosis
of the disease [
3
]. These complicated factors have raised many challenges for medical
professionals, especially those who are young and inexperienced [
4
]. Machine learning
is widely adopted to develop medical auxiliary diagnostic systems [
5
], which are also
known as Computer-Aided Diagnosis (CAD) systems. CAD systems are important tools
that provide disease diagnosis and prognosis [
6
,
7
]. They do not only help doctors make
quick decisions and save patients’ time but also reduce the uncomfortable experience
of patients by replacing invasive approaches [
8
]. CAD systems use a wide spectrum of
machine learning methods [
9
], ranging from single prediction models such as Support
Vector Machine (SVM) and Decision Tree (DT), to ensemble and deep learning models, such
Diagnostics 2021, 11, 2288. https://doi.org/10.3390/diagnostics11122288 https://www.mdpi.com/journal/diagnostics
