Precision medicine requires drug repurposing methods that adapt to individual patient profiles while working within regulatory frameworks. Existing approaches apply uniform models to all patients, only using individual factors as inputs or filters. Our framework instead integrates patient-specific profiles into the learning algorithm through a customized loss function. We combine standard link prediction with UK Biobank data—integrating polygenic risk scores, biomarker expressions, and medical history. Evaluated on a biomedical knowledge graph connecting 61,000+ entities through 1.2+ million relations, our approach improves drug repurposing quality with AUPRC improvements ranging from 1.3 × to 5.4 × across patients. Case studies on Alzheimer’s Disease patients reveal drug candidates with stronger AD evidence and patient-specific mechanisms. Our loss function identifies influential diseases and biomarkers for each patient, enhancing interpretability while providing biologically relevant recommendations tailored to individual profiles. This approach represents a fundamental shift from treating personalization as data preprocessing to embedding it within the learning objective itself. • Algorithm-level personalization. We embed patient context directly into the learning objective via a patient-specific loss that blends standard link prediction with terms guided by polygenic risk scores (PRS) and protein-biomarker deviations, optimizing for an individual rather than the population. • Clinically grounded signals. We integrate UK Biobank–derived PRS, protein biomarker levels, and diagnosis history to tailor drug–disease scores to each patient’s biology and clinical context, anchoring personalization in routinely collectable, real-world data. • Preserved generalization with better rankings. We maintain foundation-model link-prediction quality while substantially improving patient-specific drug repurposing performance (e.g., AUPRC improvements ranging from 1.3 × to 5.4 × across patients), demonstrating effectiveness without sacrificing global metrics. • Interpretability at the patient level. We learn sparse, patient-level weights over diseases and biomarkers that reveal which comorbidities and dysregulated proteins drive recommendations, supporting transparent, clinician-facing interpretation.

OBJECTIVE: Drug repurposing accelerates therapeutic development by finding new indications for approved drugs. However, accounting for individual patient differences is challenging. This study introduces a Precision Drug Repurposing (PDR) framework at single-patient resolution, integrating individual-level data with a foundational biomedical knowledge graph to enable personalized drug discovery. METHODS: We developed a framework integrating patient-specific data from the UK Biobank (Polygenic Risk Scores, biomarker expressions, and medical history) with a comprehensive biomedical knowledge graph (61,146 entities, 1,246,726 relations). Using Alzheimer's Disease as a case study, we compared three diverse patient-specific models with a foundational model through standard link prediction metrics. We evaluated top predicted candidate drugs using patient medication history and literature review. RESULTS: Our framework maintained the robust prediction capabilities of the foundational model. The integration of patient data, particularly Polygenic Risk Scores (PRS), significantly influenced drug prioritization (Cohen's d = 1.05 for scoring differences). Ablation studies demonstrated PRS's crucial role, with effect size decreasing to 0.77 upon removal. Each patient model identified novel drug candidates that were missed by the foundational model but showed therapeutic relevance when evaluated using patient's own medication history. These candidates were further supported by aligned literature evidence with the patient-level genetic risk profiles based on PRS. CONCLUSION: This exploratory study demonstrates a promising approach to precision drug repurposing by integrating patient-specific data with a foundational knowledge graph.

OBJECTIVE: Drug repurposing offers a cost-effective strategy to accelerate drug development by identifying new therapeutic uses for approved medications. Knowledge graphs (KGs) that capture large amounts of biomedical knowledge have recently been used for drug repurposing, however, KGs are inherently incomplete due to our limited biomedical knowledge. METHODS: We propose KGiA, an inductive graph augmentation method that supports semi-inductive reasoning-allowing models to generalize to previously unseen biomedical entities. KGiA enhances KGs using counterfactual relationships mined from disease-specific topological patterns. We apply it to a state-of-art biomedical KG constructed from six datasets including biomedical relationships extracted from biomedical literature, which comprised 1,614,801 triples and 100,563 entities, including 30,006 diseases. RESULTS: Across five augmented architectures, KGiA improves generalizability by up to 24×in Mean Reciprocal Rank (MRR) and outperforms the state-of-the-art KG-based drug repurposing model by up to 32%. We applied KGiA in four case studies of diseases including Alzheimer's Disease and showed its promise in identifying novel repurposed candidate drugs. CONCLUSION: We showed that leveraging counterfactual relationships derived from disease-specific graph structures to augment existing knowledge graphs improved performance in KG-based drug repurposing.

In this paper, a novel framework for cluster detection in co-operative wireless networks is proposed. This framework is modeled by a dynamic game with incomplete information, in which each player in the game aspires to improve its position in the network by forming cooperative groups. Instead of static systems, the attention we paid in this paper is highly dynamic networks, where the users' high mobility brings a huge challenge in clustering. In order to mitigate that impact, this paper is from the perspective of social relations to clustering, extracting users' social nature from their mobile patterns and designing distributed cluster strategy based on game model. The introduction of social nature with generally long-term characteristics makes clustering framework more predictive and stable. Simulations on real-world networks show that the proposed approach performs well in clustering in cooperative wireless networks.

Meeting the stringent latency requirements for delay-sensitive scenarios is a challenge in 5G Networks. Fog computing can reduce the end to end latency considerably by extending data processing capabilities to the network edge. However, the “soft real-time” characteristic of fog-based 5G networks may intensify the effect of long-tail on latency-critical scenarios, which are very sensitive to tail latency. To alleviate the long-tail latency, an online resource allocation algorithm with low complexity is proposed in this paper. With this algorithm, latency based dynamic resource allocation can be achieved through the cooperation of monitor and orchestrator of fog controllers. The simulation results show the effectiveness of this algorithm on latency optimization, tail latency reduction and resource utilization improvement.

Existing traffic engineering strategies can achieve locally optimized network capability by uniformly leading flows to candidate paths. From a global network view, software defined network (SDN) allows a central controller to steer traffic effectively, and to improve network performance significantly, both for backbone and data center networks (DCNs). However, rational using of limited or expensive resources is very important in SDN, e.g., ternary content addressable memory (TCAM). In this paper, we focus on satisfying global network objectives such as delay and packet loss, with a comprehensive consideration of other criteria (network throughput and forwarding table size). We present a flexible multi-path routing scheme and formulate this as a mixed integer linear programming. It's turned out by simulation results and analysis that the delay and packet loss performance using proposed global algorithm outperforms other existing algorithms.