Approaches and Challenges in Network Architecture For Healthcare

Approaches and Challenges in Network Architecture For Healthcare

by Leo Petersen-Khmelnitski

There are three primary types of network architecture employed to coordinate the exchange of health information across entities:

  • Centralised: patient data is collected and stored in a centralised repository, data lake, data warehouse or other databases. The repository is usually responsible to ensure data exchange, and thus is given full control over the data, including functions to authenticate, authorise and record transactions among participants to health data exchange.
  • Federated (decentralised): interconnected network of independent databases allows for data sharing and exchange. Users are granted access to the information upon request, requests may be authorised by means of automated governance mechanisms employed in decentralised, blockchain based systems.
  • Hybrid: combines variations of federated and centralised architectures to use advantages of both systems.

It also allows to combine computational capabilities of edge and cloud to include healthcare data from IoT devices. This approach is becoming more common.


Let’s take a look at a blueprint of a federated digital infrastructure, its design combines five layers:

  • National health registry to serve as a single source of health data for the nation;
  • Coverage and claims platform to serve as building blocks for large health protection schemes, allows for the horizontal and vertical expansion of schemes by Indian states, with significant resources dedicated to fraud detection;
  • Federated personal health records framework to ensure access to their own health data by patients, and access to health data by medical research;
  • National health analytics platform that combines information on multiple health initiatives, and provides a holistic view to policy makers. The platform is equipped with predictive analytics;
  • Other horizontal components, such a unique digital health ID, health data dictionaries and supply chain management for drugs, payment gateways.

The last layer is based on healthcare APIs. They outline pre-defined specifications to allow one application to use data and even functionality of another application, hence to build an ecosystem of applications that are based on one or several backbones. APIs will play an important role to ensure interoperability for person centered healthcare.

Federated approach to health related digital infrastructure is being implemented in various ways by nations worldwide, with following key features:

  • a set of architectural principles
  • a five-layered system of architectural building blocks
  • a unique health ID
  • privacy and consent management
  • national portability
  • electronic health records
  • applicable standards and regulations
  • health analytics
  • multiple access channels


These key elements listed above enable health related organisations:

  • To deploy properly – most effective deployments are close to end users, that is to patients, at locations relevant to them, to achieve real time or near real time operation. Currently, a combination of public clouds and permissioned networks results in the most efficient hybrid architectures. It may change in the future.
  • To interact with right partners – digital infrastructure allows healthcare providers to deploy an ecosystem of technology partners, service providers, regional and local specialists, as well as IT startups.
  • To be open to new opportunities and developments – the aim is to ensure a secure, private, real time, direct connection to multiple participants to the ecosystem


The most popular misguided perception among health professionals towards digital infrastructure is that it does not impact patient care. Hence, the low priority to digital infrastructure in procurement programmes.

Lack of network connectivity

The reality is that many hospitals and healthcare institutions lack the network connectivity necessary to drive technological clinical innovation. Lack of sufficient network coverage is the single most prevalent roadblock to the deployment of interconnected medical devices and other smarts.

Poor network infrastructure may also result in increased operational costs. Expansion of bandwidth does not help, as it drives up the long term costs but does not solve the underlying issues.

In order to support new AI algorithms and IoMT devices, an adequate supply of underlying computing resources is required for the training of these complicated algorithms, and to ensure permanent reliable data feeds to and from the IoMT devices. Due to the sheer size of training data as well as the nature of the deep learning models, a powerful deep learning training rig can drive down the total cost of AI product ownership.

Scanning Paper

Many healthcare institutions still use paper-based processes in the delivery of core services, wasting time and money on unnecessary administration. Some have shifted to scanning papers, even employ digital signatures, but did not alter their business processes or reform the underlying infrastructure. 

But the traditional hospital infrastructure was not designed to meet the demands of digital business. The ability to provide innovative services and better patient outcomes is underpinned by a strong foundation, the digital infrastructure.

If there is no such foundation, the inefficiencies that arise upon introduction of individual digital processes actually reduce the number of patients seen in a day and the amount of time spent with each one, increasing the likelihood that something is missed.

Too Much of IoMT 

Too many healthcare providers install multiple sensors before they figure out how they are going to use, let alone monetise the data that these sensors collect. Data minimalism is an approach to study and apply where it is deemed adequate, not only with regards to data, but with regards to hardware as well.

Contact our experts to learn how we can help your organisation