Real-time monitoring and data collection have been made possible by the proliferation of Internet of Things (IoT) devices in healthcare, which has revolutionized patient care. However, this increased connectivity also introduces significant security vulnerabilities, necessitating robust intrusion detection systems (IDSs) to protect sensitive health information. This paper proposes an advanced intrusion detection system specifically designed for IoT healthcare devices, leveraging Convolutional Neural Networks (CNNs). Our approach exploits CNN’s superior capability in feature selection by crow search algorithm to identify and mitigate unauthorized access attempts. The IDS framework is trained on NSL-KDD dataset, encompassing various attack vectors pertinent to healthcare IoT environments. Experimental results demonstrate that the CNN-based IDS achieves high accuracy of 98.5% and low false positive rates in detecting intrusions, outperforming traditional machine learning models. The proposed system enhances the security posture of IoT healthcare networks, ensuring reliable and secure patient data management. This research highlights the critical importance of integrating advanced neural network techniques in developing effective cybersecurity solutions for IoT in healthcare.