A Smart Healthcare Recommendation System for Multidisciplinary Diabetes Patients with Data Fusion Based on Deep Ensemble Learning

Table 1

Summary of existing literature reviews.

Paper	Classification methodology adopted	Limitations	Advantages
[12]	ML	1. Single dataset; 2. No data fusion; 3. Only structured data	Only the optimal feature selection technique was adopted
[13]	ML and AI	1. Single dataset; 2. No data fusion; 3. Only DR image data	The bloodless technique was adopted
[18]	DML and generalized linear model	1. Single EHRs dataset; 2. No data fusion	1. Electronic health record; 2. Data fusion; 3. Feature selection
[19]	AI	1. Single dataset; 2. No data fusion; 3.Only DR image data	1. Automated software; 2. Smartphone-based DR and sight-threatening detection
[20]	AI and ML	1. Single dataset; 2. No data fusion	Incorporating wearable devices and IoT to collect and manage big data
[29]	Supervised ML	1. Single dataset; 2. No data fusion	1. Combine structured and unstructured data; 2. Feature selection