Enhancing Federated Learning for Confidential Sensor Data Aggregation in IoMT Environments Dagmawit Tadesse Aga Dept. of Computer Science North Carolina A&T State University Greensboro, USA daga@aggies.ncat.edu Madhuri Siddula Dept. of computer science North Carolina A&T State University Greensboro, USA msiddula@ncat.edu Abstract—The Internet of Medical Things (IoMT) is a trans- formative technology that enables medical systems and devices to collaborate seamlessly to improve healthcare delivery. However, the widespread adoption of IoMT raises significant privacy concerns, particularly regarding the aggregation and analysis of sensitive medical data. This paper proposes a novel approach to address these challenges through the utilization of Feder- ated Learning (FL) techniques. We present a comprehensive framework for privacy-preserving data aggregation, leveraging FL to collaboratively train machine learning models across distributed devices while preserving data privacy and security. Our approach decentralizes the model training process and performs computations locally on edge devices. This ensures that sensitive patient data remains secure and never leaves the respective devices, eliminating the need to share data with a central server. The central server combines the weights of the parameters and transmits only the updated weights to each device. Furthermore, we conducted experimental evaluations to demonstrate the effectiveness and efficiency of our proposed approach by achieving high accuracy (90.91%) while ensuring privacy. Additionally, we compared the model with other models and related work to evaluate its performance in terms of accu- racy, privacy preservation, and overall effectiveness in securing IoMT data. Overall, our work contributes to the advancement of privacy-enhancing technologies for IoMT, paving the way for more secure and trustworthy healthcare systems in the era of connected medical devices. Index Terms—Data aggregation, Data privacy, Internet of Things, Internet of Medical Things, TensorFlow Federated (TFF), Federated Learning I. I NTRODUCTION The Internet of Medical Things (IoMT) is a rapidly grow- ing field that refers to the network of medical devices and applications that connect to healthcare information technology systems through online computer networks [23]. These devices collect and transmit patient data, which can then be used to improve diagnosis, treatment, and overall healthcare outcomes [24]. IoMT has the potential to revolutionize healthcare by improving patient care, reducing costs, and making healthcare more efficient [25]. Around 60% of healthcare enterprises and medical establishments have already integrated IoT (IoMT), resulting significant cost savings, increased profitability, and enhanced customer experience [11]. The global market size for IoMT was 41.17 billion dollars in 2020, and it is anticipated to rise from 30.79 billion dollars in 2021 to 187.60 billion dollars in 2028 [20]. However, it is important to address the challenges associated with IoMT to ensure its safe and effective implementation [9]. The growing use of healthcare data in the IoMT and other digital health solutions brings immense benefits but also raises significant privacy challenges. It is important to ensure that IoMT devices are secure and that patient data is protected. IoMT devices collect and transmit sensitive patient data, which raises concerns about security and privacy [13]. A survey made on health organizations in the US as of 2022 shows that 21% of them reported they spend 500 thousand to 1 million US Dollars on IoT/IoMT device security. Additionally, 21% reported spending between one and 2.5 million dollars on device security. [22]. Fig. 1. Annual U.S. healthcare sector IoT/IoMT device security spend 2022 By understanding the challenges and available techniques, researchers and healthcare professionals can work towards responsible and secure data aggregation that unlocks valuable insights while safeguarding patient privacy. Privacy-preserving data aggregation for healthcare data requires techniques that allow for the collection and analysis of valuable insights from the data while ensuring individual patient information remains confidential [27]. Various techniques are being developed to address these concerns while still enabling valuable data aggre- gation for analysis and insights. Thankfully, various techniques are being developed to address this delicate balance. Some of these techniques that have been utilized to preserve healthcare data include differential privacy, which adds noise to the data