On-Device Interpretable Tsetlin Machine-Based Intrusion Detection for Secure IoMT PubID: pubid: 979-8-3315-1276-8/26/$31.00 ©2026 IEEE
Rahul Jaiswal, Per-Arne Andersen, Linga Reddy Cenkeramaddi, Lei Jiao, Ole-Christoffer Granmo The Centre for Artificial Intelligence Research (CAIR) {rahul.jaiswal, per.andersen, linga.cenkeramaddi, lei.jiao, ole.granmo}@uia.no · 2026
Read On-Device Interpretable Tsetlin Machine-Based Intrusion Detection for Secure IoMT PubID: pubid: 979-8-3315-1276-8/26/$31.00 ©2026 IEEE by Rahul Jaiswal, Per-Arne Andersen, Linga Reddy Cenkeramaddi, Lei Jiao, Ole-Christoffer Granmo The Centre for Artificial Intelligence Research (CAIR) {rahul.jaiswal, per.andersen, linga.cenkeramaddi, lei.jiao, ole.granmo}@uia.no (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-16707.txt.
Introduction
Jaiswal and colleagues propose an on-device, interpretable intrusion-detection system for Internet of Medical Things networks, built on Tsetlin Machines. The rule-driven model represents attack patterns in propositional logic and explains decisions through feature contributions, vote scores, and clause heatmaps. On the MedSec-25 dataset covering realistic attack phases, it reaches a macro F1 score of 97.83%, beating standard machine-learning models. Raspberry Pi deployment supports real-time on-device inference.
Catalog record
| CREATOR | Rahul Jaiswal, Per-Arne Andersen, Linga Reddy Cenkeramaddi, Lei Jiao, Ole-Christoffer Granmo The Centre for Artificial Intelligence Research (CAIR) {rahul.jaiswal, per.andersen, linga.cenkeramaddi, lei.jiao, ole.granmo}@uia.no |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36745 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.16707
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On-Device Interpretable Tsetlin Machine-Based Intrusion Detection for Secure IoMT PubID: pubid: 979-8-3315-1276-8/26/$31.00 ©2026 IEEE Rahul Jaiswal, Per-Arne Andersen, Linga Reddy Cenkeramaddi, Lei Jiao, Ole-Christoffer Granmo The Centre for Artificial Intelligence Research (CAIR) {rahul.jaiswal, per.andersen, linga.cenkeramaddi, lei.jiao, ole.granmo}@uia.no Abstract The rapid evolution of digital health technologies is redefining healthcare services worldwide. The integration of wireless communication and Internet-enabled medical devices within Internet of Medical Things (IoMT) networks enables continuous, real-time patient monitoring. However, this increased connectivity raises cybersecurity and patient safety risks due to increasingly sophisticated cyberattacks. This paper proposes a novel on-device, interpretable Tsetlin Machine (TM)-based Intrusion Detection System (IDS) to identify various phases of cyberattacks in IoMT environments. The TM is a rule-driven and transparent machine learning (ML) approach that represents attack patterns using propositional logic. Extensive evaluations on the MedSec-25 dataset, encompassing various phases of realistic cyberattacks, show that the proposed model outperforms ML models and state-of-the-art methods, attaining a classification performance of 97.83%. Moreover, the proposed model offers explicit explanations of its decisions to enhance transparency using feature-level contributions, class-wise vote scores, and clause activation heatmaps. Edge deployment (Raspberry Pi) further supports real-time on-device inference and intrusion detection. The combination of interpretability and high performance makes the proposed model well-suited for IoMT healthcare, where trust, reliability, safety, and timely decision-making are critical. I Introduction The digital health technologies are advancing rapidly, enabling easier access to online doctor consultations through better internet connectivity, smartphones, and mobile applications. This approach is not only faster and more affordable for busy professionals but also highly convenient for elderly people, especially those living in rural or remote areas. It eliminates the need for unnecessary travel to the hospital, reduces waiting time, and offers a cost-effective alternative to traditional in-person doctor visits. In digital healthcare services, patients’ health data is transmitted electronically to the healthcare providers, such as hospitals and doctors, via digital channels using wearable medical devices (e.g., smart watches), as shown in Fig. 1 . The interconnected framework that enables medical devices, healthcare platforms, and digital systems to exchange medical data over the Internet is known as the Internet of Medical Things (IoMT) [ 16 ] . It enables efficient transmission of health data from patient devices to healthcare providers, improving both the timeliness and effectiveness of ongoing patient care. Fig. 1: Overview of IoMT in healthcare. The Deloitte Health Care Report 2026 [ 12 ] highlights that IoMT and digital health technologies are driving healthcare toward a “Care Anywhere” model, making virtual healthcare services more accessible and efficient. However, the use of IoMT medical devices introduces several critical challenges. Since these devices are connected to the Internet and continuously transmit highly sensitive medical data, they become attractive targets for cyberattacks. This raises major concerns such as unauthorized data access, medical device hijacking, exposure of sensitive personal health information, and cyberattacks that may disrupt system functionality or render medical devices unavailable, ultimately posing risks to both patient safety and privacy. For instance, if the cyberattackers alter the medical data of the patient within the hospital systems, then doctors may rely on the inaccurate readings, resulting in inappropriate treatment and potentially life-threatening harm to the patients. The CrowdStrike Global Threat Report 2026 [ 10 ] highlights that 10% of global cyberattacks target healthcare systems. To ensure safe, secure, and reliable healthcare services, Intrusion Detection Systems (IDS) [ 13 ] are used in the IoMT networks. They utilize network-based security techniques to monitor network traffic and identify potential cyberattacks in real time. Upon detecting suspicious activities, IDS promptly alerts the network administrators about potential intrusions or cyberattacks, enabling a timely response and mitigation. This paper presents a novel IDS system for detecting different phases of cyberattacks targeting IoMT environments by leveraging an interpretable machine learning (ML) approach based on the Tsetlin Machine (TM) [ 25 , 14 ] . Building on this, network-level features are extracted from IoMT traffic and used as input to the TM model, which classifies the traffic as benign (normal) or malicious attacks . To promote transparency and build trust in the outcome of the TM model beyond black-box ML models, the interpretability of the TM model is explicitly analyzed. Moreover, the proposed TM-based IDS is deployed on the resource-constrained edge device (Raspberry Pi) to enable on-device inference for real-time attack/intrusion detection. The rest of this paper is structured as follows. Section II reviews the related work and outlines the motivation. Section III describes the classifiers employed and the on-device inference approach used in this study. Section IV describes the proposed TM-based intrusion detection system. Section V introduces the experimental dataset. Section VI presents and discusses the results. Finally, Section VII concludes the paper and highlights directions for future research. II Related Work and Motivation Intrusion Detection Systems (IDS) play a critical role in safeguarding sensitive medical data from cyberattacks within IoMT environments. It can detect both internal and external attacks that bypass existing security measures. Traditional IDS systems are broadly categorized into two approaches, namely, signature-based and anomaly-based detection [ 28 , 7 ] . The signature-based approach identifies known attacks by matching observed patterns against predefined signatures. In contrast, anomaly-based detection focuses on identifying deviations from the normal network behavior to detect potential attacks. However, the highly dynamic nature of IoMT environments constrains the effectiveness of both approaches in identifying new and unseen cyberattacks, leaving IoMT devices exposed to emerging cyberattacks. Machine learning (ML) and deep learning (DL) based intrusion detection systems have demonstrated strong potential for enhancing IoMT security. These techniques can identify attacks by detecting deviations in the attributes of network traffic. For instance, the work in [ 4 ] investigates various ML approaches for cyberattack detection on the IEEE DataPort dataset and attains an accuracy of 89.89% for binary classification using K-nearest Neighbours (KNN). In [ 5 ] , a deep autoencoder-based IDS is developed to secure IoMT using the NF-ToN-IoT dataset, achieving an accuracy of 89% for 10-class classification. In [ 23 ] , the authors propose deep neural network (DNN) and long short-term memory (LSTM) based models for cyberattack detection in IoMT environments using the CICIoMT24 dataset, achieving accuracies of 78% and 79%, respectively, for six-class classification. In [ 11 ] , the authors apply various ML techniques for cyberattack detection, achieving an accuracy of 73.5% in a six-class classification setting. Although these ML/DL models can attain high classification accuracy, interpretability remains a critical requirement for their effective adoption in healthcare services. Moreover, many of these models operate as black boxes, providing limited insight into their decision-making processes, which raises significant concerns regarding the transparency of outcomes. The Tsetlin Machine (TM) [ 25 , 14 ] has emerged as a promising alternative to address these limitations. The TM is an emerging machine learning technique based on Tsetlin Automata [ 14 ] , well-suited for IoMT intrusion detection. It focuses on identifying recurring patterns in network traffic through frequent pattern learning and efficient resource allocation, rather than relying solely on error minimization. By representing knowledge as interpretable conjunctive clauses, the TM reduces overfitting and simplifies intrusion detection into manageable patterns, making it suitable for reliable on-device intrusion detection. For instance, the study in [ 15 ] introduces a TM-based anomaly detection framework and