The rapid adoption of Internet of Things (IoT)–enabled drug delivery networks, such as automated insulin pumps and smart IV infusion systems, has significantly improved precision and continuity of patient care. However, these cyber-physical medical devices are highly vulnerable to cyber-attacks including data spoofing, command injection, denial-of-service, and replay attacks, which can manipulate dosage delivery and pose life-threatening risks to patients. This paper addresses the critical challenge of detecting cyber-attacks in real time within IoT-based drug delivery environments while maintaining low computational overhead and minimal response delay.
To overcome these challenges, a hybrid deep learning–based cyber-attack detection framework using Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks is proposed. The CNN component effectively extracts spatial features from network traffic and device telemetry data, while the LSTM module captures temporal attack patterns and sequential anomalies. The proposed model is evaluated on simulated IoT medical network traffic representing normal and attack scenarios.
Experimental results demonstrate a detection accuracy of 98.7%, a low false alarm rate of 1.9%, and an average detection latency of under 120 ms, making it suitable for real-time medical response systems. The findings confirm that the proposed approach significantly enhances the cybersecurity resilience of IoT-based drug delivery networks without compromising patient safety or system performance.
Keywords: IoT Security, Cyber-Attack Detection, Smart Drug Delivery Systems, Insulin Pumps, IV Infusion Networks, CNN-LSTM, Medical Cyber-Physical Systems, Real-Time Anomaly Detection.
How to cite this article: Sruthi MV, Meruva S, Prasannakumar M, Ravikanth S, Khan PI, Kumar BS. Cyber-attack detection in IoT-based drug delivery networks. Int J Drug Deliv Technol. 2026;16(3s): 878-883; DOI: 10.25258/ijddt.16.3s.107
Source of support: Nil.
Conflict of interest: None