Table 1 Comparative analysis of existing...

Table 1

Comparative analysis of existing methods in Mpox prediction and their challenges

Related work	Year	Dataset	Classification	Algorithm	Contribution	Unaddressed challenges/Reasons for limitation
Campana, Colussi, Delmastro, Mascetti, and Pagani (2024)	2024	MCSI	Multiclass	MobileNetV3	Introduced new dataset of skin images (lesion/non-lesion) for Mpox detection	Limited optimization for edge deployment and latency-sensitive applications
Ahsan et al. (2024)	2024	MDS22	Multiclass	M-ResNet50	Investigated Mpox diagnosis using CNN with transfer and federated learning	Insufficient labeling diversity and class ambiguity hinder generalization
Ahsan et al. (2023)	2023	MDS22	Multiclass	Xception	Applied GRA-TLA for optimization and CNN enhancement	Inability to capture fine-grained lesion variations and spatial dependencies
Bala et al. (2023)	2023	MSID	Multiclass	M-DenseNet201	Proposed MonkeyNet for Mpox image classification	Failure to emphasize discriminative lesion regions reduces precision
Nayak et al. (2023)	2023	MSLD	Binary	M-ResNet18	Detection of Mpox lesions using deep learning	Poor performance under racial and illumination variations
Rabie and Saleh (2023)	2023	MPX-Data, MPPD	Multiclass/Binary	SNB, LKNN, DLC	Ensemble classification for integrated AMDS Mpox system	Lack of deep or hybrid ensemble models limits robustness
Saleh and Rabie (2023)	2023	MPX-Data	Binary	IBCO, WNB, WKNN, LSTM	Human Mpox classification using hybrid traditional methods	Non-scalable for mobile or real-time deployment
Sahin, Oztel, and Yolcu Oztel (2022)	2022	MSLD	Binary	MobileNetV2	Human Mpox classification with transfer learning	Fails to benchmark against state-of-the-art CNNs or transformers
Sitaula and Shahi (2022)	2022	MDS22	Multiclass	Comparative DL models	Compared multiple pre-trained models for Mpox	Inefficient inference and high computational demand
Thieme et al. (2023)	2023	MPXV	Binary	CNN	Binary Mpox detection using CNNs	Simple CNN backbone limits adaptability to diverse lesion types
Yadav and Qidwai (2024)	2024	MSLD	Binary	M-ResNet50 + MXGBoost	RN-50-ZCA model enhances feature extraction	Lack of regularization and dataset diversity causing overfitting
Asif, Zhao, Tang, Zhu, and Zhao (2023)	2023	MSLD	Binary	DenseNet201, MobileNet, DenseNet169	MO-WAE model integrating metaheuristic optimization	Unbalanced data distribution affects reproducibility and model stability

Related work	Year	Dataset	Classification	Algorithm	Contribution	Unaddressed challenges/Reasons for limitation
Campana, Colussi, Delmastro, Mascetti, and Pagani (2024)	2024	MCSI	Multiclass	MobileNetV3	Introduced new dataset of skin images (lesion/non-lesion) for Mpox detection	Limited optimization for edge deployment and latency-sensitive applications
Ahsan et al. (2024)	2024	MDS22	Multiclass	M-ResNet50	Investigated Mpox diagnosis using CNN with transfer and federated learning	Insufficient labeling diversity and class ambiguity hinder generalization
Ahsan et al. (2023)	2023	MDS22	Multiclass	Xception	Applied GRA-TLA for optimization and CNN enhancement	Inability to capture fine-grained lesion variations and spatial dependencies
Bala et al. (2023)	2023	MSID	Multiclass	M-DenseNet201	Proposed MonkeyNet for Mpox image classification	Failure to emphasize discriminative lesion regions reduces precision
Nayak et al. (2023)	2023	MSLD	Binary	M-ResNet18	Detection of Mpox lesions using deep learning	Poor performance under racial and illumination variations
Rabie and Saleh (2023)	2023	MPX-Data, MPPD	Multiclass/Binary	SNB, LKNN, DLC	Ensemble classification for integrated AMDS Mpox system	Lack of deep or hybrid ensemble models limits robustness
Saleh and Rabie (2023)	2023	MPX-Data	Binary	IBCO, WNB, WKNN, LSTM	Human Mpox classification using hybrid traditional methods	Non-scalable for mobile or real-time deployment
Sahin, Oztel, and Yolcu Oztel (2022)	2022	MSLD	Binary	MobileNetV2	Human Mpox classification with transfer learning	Fails to benchmark against state-of-the-art CNNs or transformers
Sitaula and Shahi (2022)	2022	MDS22	Multiclass	Comparative DL models	Compared multiple pre-trained models for Mpox	Inefficient inference and high computational demand
Thieme et al. (2023)	2023	MPXV	Binary	CNN	Binary Mpox detection using CNNs	Simple CNN backbone limits adaptability to diverse lesion types
Yadav and Qidwai (2024)	2024	MSLD	Binary	M-ResNet50 + MXGBoost	RN-50-ZCA model enhances feature extraction	Lack of regularization and dataset diversity causing overfitting
Asif, Zhao, Tang, Zhu, and Zhao (2023)	2023	MSLD	Binary	DenseNet201, MobileNet, DenseNet169	MO-WAE model integrating metaheuristic optimization	Unbalanced data distribution affects reproducibility and model stability

Note(s): MCSI: Mpox Close Skin Images; MDS22: Monkeypoxdataset-2022; MPXV: Monkeypox Virus; MPX-Data: Monkeypox-Data; MSLD: Monkeypox Skin Lesion Dataset; MSID: Monkeypox Skin Images Dataset; M-DenseNet201: Modified DenseNet-201; M-ResNet18: Modified ResNet-18; M-XGBoost: Modified XGBoost; M-ResNet50: Modified ResNet50; CNN: Convolutional Neural Network; LSTM: Long Short-Term Memory; WKNN: Weighted K-Nearest Neighbors; WNB: Weighted Naïve Bayes; IBCO: Improved Binary Chimp Optimization; AMDS: Accurate Monkeypox Diagnosing Strategy; LKNN: Layered K-Nearest Neighbors; DLC: Deep Learning Classifier; GRA-TLA: Generalization and Regularization-based Transfer Learning; MO-WAE: Metaheuristic-Optimization-based Weighted Average Ensemble; RN-50-ZCA: Residual Network-50-Zero Phase Component Analysis

[ViewLarge]

Sharing Unavailable