Federated and distributed learning applications for electronic health records and structured medical data: a scoping review

OBJECTIVES

Federated learning (FL) has gained popularity in clinical research in recent years to facilitate privacy-preserving collaboration. Structured data, one of the most prevalent forms of clinical data, has experienced significant growth in volume concurrently, notably with the widespread adoption of electronic health records in clinical practice. This review examines FL applications on structured medical data, identifies contemporary limitations, and discusses potential innovations.

MATERIALS AND METHODS

We searched 5 databases, SCOPUS, MEDLINE, Web of Science, Embase, and CINAHL, to identify articles that applied FL to structured medical data and reported results following the PRISMA guidelines. Each selected publication was evaluated from 3 primary perspectives, including data quality, modeling strategies, and FL frameworks.

RESULTS

Out of the 1193 papers screened, 34 met the inclusion criteria, with each article consisting of one or more studies that used FL to handle structured clinical/medical data. Of these, 24 utilized data acquired from electronic health records, with clinical predictions and association studies being the most common clinical research tasks that FL was applied to. Only one article exclusively explored the vertical FL setting, while the remaining 33 explored the horizontal FL setting, with only 14 discussing comparisons between single-site (local) and FL (global) analysis.

CONCLUSIONS

The existing FL applications on structured medical data lack sufficient evaluations of clinically meaningful benefits, particularly when compared to single-site analyses. Therefore, it is crucial for future FL applications to prioritize clinical motivations and develop designs and methodologies that can effectively support and aid clinical practice and research.

Distributed algorithms from arboreal ants for the shortest path problem

Colonies of the arboreal turtle ant create networks of trails that link nests and food sources on the graph formed by branches and vines in the canopy of the tropical forest. Ants put down a volatile pheromone on the edges as they traverse them. At each vertex, the next edge to traverse is chosen using a decision rule based on the current pheromone level. There is a bidirectional flow of ants around the network. In a previous field study, it was observed that the trail networks approximately minimize the number of vertices, thus solving a variant of the popular shortest path problem without any central control and with minimal computational resources. We propose a biologically plausible model, based on a variant of the reinforced random walk on a graph, which explains this observation and suggests surprising algorithms for the shortest path problem and its variants. Through simulations and analysis, we show that when the rate of flow of ants does not change, the dynamics converges to the path with the minimum number of vertices, as observed in the field. The dynamics converges to the shortest path when the rate of flow increases with time, so the colony can solve the shortest path problem merely by increasing the flow rate. We also show that to guarantee convergence to the shortest path, bidirectional flow and a decision rule dividing the flow in proportion to the pheromone level are necessary, but convergence to approximately short paths is possible with other decision rules.

The CODYSUN Approach: A Novel Distributed Paradigm for Dynamic Spectrum Sharing in Satellite Communications

With a constant increase in the number of deployed satellites, it is expected that the current fixed spectrum allocation in satellite communications (SATCOM) will migrate towards more dynamic and flexible spectrum sharing rules. This migration is accelerated due to the introduction of new terrestrial services in bands used by satellite services. Therefore, it is important to design dynamic spectrum sharing (DSS) solutions that can maximize spectrum utilization and support coexistence between a high number of satellite and terrestrial networks operating in the same spectrum bands. Several DSS solutions for SATCOM exist, however, they are mainly centralized solutions and might lead to scalability issues with increasing satellite density. This paper describes two distributed DSS techniques for efficient spectrum sharing across multiple satellite systems (geostationary and non-geostationary satellites with earth stations in motion) and terrestrial networks, with a focus on increasing spectrum utilization and minimizing the impact of interference between satellite and terrestrial segments. Two relevant SATCOM use cases have been selected for dynamic spectrum sharing: the opportunistic sharing of satellite and terrestrial systems in (i) downlink Ka-band and (ii) uplink Ka-band. For the two selected use cases, the performance of proposed DSS techniques has been analyzed and compared to static spectrum allocation. Notable performance gains have been obtained.

Toward Accelerated Training of Parallel Support Vector Machines Based on Voronoi Diagrams

Typical applications of wireless sensor networks (WSN), such as in Industry 4.0 and smart cities, involves acquiring and processing large amounts of data in federated systems. Important challenges arise for machine learning algorithms in this scenario, such as reducing energy consumption and minimizing data exchange between devices in different zones. This paper introduces a novel method for accelerated training of parallel Support Vector Machines (pSVMs), based on ensembles, tailored to these kinds of problems. To achieve this, the training set is split into several Voronoi regions. These regions are small enough to permit faster parallel training of SVMs, reducing computational payload. Results from experiments comparing the proposed method with a single SVM and a standard ensemble of SVMs demonstrate that this approach can provide comparable performance while limiting the number of regions required to solve classification tasks. These advantages facilitate the development of energy-efficient policies in WSN.

CoNTe: A Core Network Temporal Blockchain for 5G

Virtual Network Functions allow the effective separation between hardware and network functionality, a strong paradigm shift from previously tightly integrated monolithic, vendor, and technology dependent deployments. In this virtualized paradigm, all aspects of network operations can be made to deploy on demand, dynamically scale, as well as be shared and interworked in ways that mirror behaviors of general cloud computing. To date, although seeing rising demand, distributed ledger technology remains largely incompatible in such elastic deployments, by its nature as functioning as an immutable record store. This work focuses on the structural incompatibility of current blockchain designs and proposes a novel, temporal blockchain design built atop federated byzantine agreement, which has the ability to dynamically scale and be packaged as a Virtual Network Function (VNF) for the 5G Core.

Average Consensus over Mobile Wireless Sensor Networks: Weight Matrix Guaranteeing Convergence without Reconfiguration of Edge Weights

Efficient data aggregation is crucial for mobile wireless sensor networks, as their resources are significantly constrained. Over recent years, the average consensus algorithm has found a wide application in this technology. In this paper, we present a weight matrix simplifying the average consensus algorithm over mobile wireless sensor networks, thereby prolonging the network lifetime as well as ensuring the proper operation of the algorithm. Our contribution results from the theorem stating how the Laplacian spectrum of an undirected simple finite graph changes in the case of adding an arbitrary edge into this graph. We identify that the mixing parameter of Best Constant weights of a complete finite graph with an arbitrary order ensures the convergence in time-varying topologies without any reconfiguration of the edge weights. The presented theorems and lemmas are verified over evolving graphs with various parameters, whereby it is demonstrated that our approach ensures the convergence of the average consensus algorithm over mobile wireless sensor networks in spite of no edge reconfiguration.

Gaussian Belief Propagation for Solving Network Utility Maximization with Delivery Contracts

Classical network utility maximization (NUM) models fail to capture network dynamics, which are of increasing importance for modeling network behaviors. In this paper, we consider the NUM with delivery contracts, which are constraints to the classical model to describe network dynamics. This paper investigates a method to distributively solve the given problem. We first transform the problem into an equivalent model of linear equations by dual decomposition theory, and then use Gaussian belief propagation algorithm to solve the equivalent issue distributively. The proposed algorithm has faster convergence speed than the existing first-order methods and distributed Newton method. Experimental results have demonstrated the effectiveness of our proposed approach.

A Message-Passing Algorithm for Wireless Network Scheduling

We consider scheduling in wireless networks and formulate it as Maximum Weighted Independent Set (MWIS) problem on a "conflict" graph that captures interference among simultaneous transmissions. We propose a novel, low-complexity, and fully distributed algorithm that yields high-quality feasible solutions. Our proposed algorithm consists of two phases, each of which requires only local information and is based on message-passing. The first phase solves a relaxation of the MWIS problem using a gradient projection method. The relaxation we consider is tighter than the simple linear programming relaxation and incorporates constraints on all cliques in the graph. The second phase of the algorithm starts from the solution of the relaxation and constructs a feasible solution to the MWIS problem. We show that our algorithm always outputs an optimal solution to the MWIS problem for perfect graphs. Simulation results compare our policies against Carrier Sense Multiple Access (CSMA) and other alternatives and show excellent performance.

A Message-Passing Algorithm for Wireless Network Scheduling

We consider scheduling in wireless networks and formulate it as Maximum Weighted Independent Set (MWIS) problem on a "conflict" graph that captures interference among simultaneous transmissions. We propose a novel, low-complexity, and fully distributed algorithm that yields high-quality feasible solutions. Our proposed algorithm consists of two phases, each of which requires only local information and is based on message-passing. The first phase solves a relaxation of the MWIS problem using a gradient projection method. The relaxation we consider is tighter than the simple linear programming relaxation and incorporates constraints on all cliques in the graph. The second phase of the algorithm starts from the solution of the relaxation and constructs a feasible solution to the MWIS problem. We show that our algorithm always outputs an optimal solution to the MWIS problem for perfect graphs. Simulation results compare our policies against Carrier Sense Multiple Access (CSMA) and other alternatives and show excellent performance.