Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jul 2;15(1):22909.
doi: 10.1038/s41598-025-05850-z.

A deep dive into artificial intelligence with enhanced optimization-based security breach detection in internet of health things enabled smart city environment

S Jayanthi  1 , Sodagudi Suhasini  2 , N Sharmili  3 , E Laxmi Lydia  4 , V Shwetha  5 , Bibhuti Bhusan Dash  6 , Mrinal Bachute  7
Affiliations

A deep dive into artificial intelligence with enhanced optimization-based security breach detection in internet of health things enabled smart city environment

S Jayanthi et al. Sci Rep. .

Abstract

Internet of Health Things (IoHT) plays a vital role in everyday routine by giving electronic healthcare services and the ability to improve patient care quality. IoHT applications and devices become widely susceptible to cyber-attacks as the tools are smaller and varied. Additionally, it is of dual significance once IoHT contains tools applied in the healthcare field. In the context of smart cities, IoHT enables proactive health management, remote diagnostics, and continuous patient monitoring. Therefore, it is essential to advance a strong cyber-attack detection method in the IoHT environments to mitigate security risks and prevent devices from being vulnerable to cyber-attacks. So, improving an intrusion detection system (IDS) for attack identification and detection using the IoHT method is fundamentally necessary. Deep learning (DL) has recently been applied in attack detection because it can remove and learn deeper features of known attacks and identify unknown attacks by analyzing network traffic for anomalous patterns. This study presents a Securing Attack Detection through Deep Belief Networks and an Advanced Metaheuristic Optimization Algorithm (SADDBN-AMOA) model in smart city-based IoHT networks. The main aim of the SADDBN-AMOA technique is to provide a resilient attack detection method in the IoHT environment of smart cities to mitigate security threats. The data pre-processing phase applies the Z-score normalization method for converting input data into a structured pattern. For the selection of the feature process, the proposed SADDBN-AMOA model designs a slime mould optimization (SMO) model to select the most related features from the data. Followed by the deep belief network (DBN) method is used for the attack classification method. Finally, the improved Harris Hawk optimization (IHHO) approach fine-tunes the hyperparameter values of the DBN method, leading to superior classification performances. The effectiveness of the SADDBN-AMOA method is investigated under the IoT healthcare security dataset. The experimental validation of the SADDBN-AMOA method illustrated a superior accuracy value of 98.71% over existing models.

Keywords: Artificial intelligence; Attack detection; Feature selection; Harris Hawk optimization; Internet of health things; Intrusion detection systems; Smart cities.

PubMed Disclaimer

Conflict of interest statement

Declarations. Competing interests: The authors declare no competing interests. Ethics approval: This article contains no studies with human participants performed by any authors.

Figures

Fig. 1
Fig. 1
General architecture for the IoHT systems.
Fig. 2
Fig. 2
Overall process of SADDBN-AMOA model.
Fig. 3
Fig. 3
DBN architecture.
Fig. 4
Fig. 4
Confusion matrix of SADDBN-AMOA model under TRPHE and TSPHE of (a-b) 80:20 (c-d) 70:30.
Fig. 5
Fig. 5
Average of SADDBN-AMOA method on 80%TRPHE and 20%TSPHE.
Fig. 6
Fig. 6
Average of SADDBN-AMOA model on 70%TRPHE and 30%TSPHE.
Fig. 7
Fig. 7
(a-c) formula image analysis on 80:20 and 70:30 and (b-d) Loss graph on 80:20 and 70:30
Fig. 8
Fig. 8
(a-c) PR graph on 80:20 and 70:30 and (b-d) ROC graph on 80:20 and 70:30.
Fig. 9
Fig. 9
Comparative study of SADDBN-AMOA method.
Fig. 10
Fig. 10
CT outcome of SADDBN-AMOA model with existing techniques.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Aruna Santhi, J. & Vijaya Saradhi, T. Attack detection in medical internet of things using optimized deep learning: Enhanced security in healthcare sector. Data Technol. Appl.55 (5), 682–714 (2021).
    1. Saheed, Y. K. & Arowolo, M. O. Efficient cyber-attack detection on the internet of medical things-smart environment based on deep recurrent neural network and machine learning algorithms. IEEe Access.9, 161546–161554 (2021).
    1. Al Abdulwahid, A. Detection of middlebox-based attacks in healthcare internet of things using multiple machine learning models. Comput. Intell. Neurosci.2022(1), 2037954 (2022). - PMC - PubMed
    1. Shankar, K. Improving the security and authentication of the cloud with IoT using hybrid optimization based quantum hash function. J. Intell. Syst. Internet Things. 1 (2), 61–61 (2021).
    1. Ravi, V., Pham, T. D. & Alazab, M. Deep learning-based network intrusion detection system for internet of medical things. IEEE Internet Things Mag. 6 (2), 50–54 (2023).

LinkOut - more resources