Taxonomy completion is the task of integrating new concepts into an existing taxonomy by determining the appropriate hypernym--hyponym relations. Existing approaches often struggle with the inherent imbalance between leaf and non-leaf edges, which induces bias in representation learning. In this paper, we propose BLEND: B alanced and L eaf- En hanced D ual Fine-Tuning for Taxonomy Completion, a novel framework designed to mitigate this inductive bias. Our method employs independent fine-tuning of two lightweight large language models (LLMs): one optimized with a leaf-focused objective and the other trained with a balanced focused strategy. To further enhance structural understanding, we apply contrastive learning over structure-encoded paths and introduce a combined loss function, enabling more robust representation of hierarchical relations. Extensive experiments on three real-world benchmark datasets demonstrate that BLEND achieves up to 9.32% improvement in recall or hit metrics compared to state-of-the-art approaches. Moreover, BLEND delivers efficient inference while outperforming the latest baseline COMI, highlighting its effectiveness for taxonomy completion tasks.

Cardiovascular disease (CVD) is one of the leading causes of the mortality rate increase. To effectively analyze wearable sensor data for providing accurate and reliable estimation of vital signs, such as heart rate (HR), the use of artificial intelligence (AI) in wearable devices is increasing. The use of AI in designing healthcare wearable sensors is crucial to the uprising of the Internet of Medical Things (IoMT). The appearance of IoMT sensor technologies lets the healthcare industry shift from vis-a-vis consulting to telemedicine. IoMT sensors have transformed the healthcare industry by improving patient safety and reducing healthcare costs. This article proposes a photoplethysmogram (PPG) enabled wearable device in an edge-IoMT computing environment that enables users to monitor their real-time health status. A deep learning approach for automatic feature extraction is proposed in this work. The deep learning algorithm learns features from a super-resolution spectrogram computed using superlet transform. In the proposed system, a PPG signal is input, and the output layer provides information on HR. The proposed framework uses two publicly available PPG data sets to train and test the proposed edge-assisted model. The model is further evaluated using an in-house acquired PPG signal data set. The proposed framework obtained a mean absolute error of 0.76, 1.01, 1.46, and 1.79 BPM for IEEE Signal Processing Cup 2015 (IEEE SPC) training, IEEE SPC test, BAMI-I, and BAMI-II data sets, respectively. The proposed edge-based IoMT framework satisfactorily predicts HR in real time using reference-signal-less PPG sensor signal.