As I have discussed before in my interoperability series, federated viewers are remarkably effective within regional contexts, beyond which scaling of the model becomes impractical. I have also discussed the need and ways of building intelligence into a federated framework. Here I am going to expand further on my previous ideas.
A typical workflow of Health Information Exchange (HIE)
A clinician needs access to patient information from another hospital information system. He initiates a search using the available patient details such as name and health card number. The request is sent to the HIE server. The HIE server initiates a search on all connected Hospital Information Systems (HIS). Each HIS responds with records matching the search. The HIE server collects all responses and presents a consolidated view to the user.
The clinician selects the right patient from a list of search responses and sends a request back to HIE server for detailed information, and the process continues. Just as a chain is no stronger than its weakest link, the HIE server response is never faster than the slowest responding HIS. Speed becomes a limiting factor as the system scales to more and more HIS and the clinician perceives it as slow, impeding the clinical workflow. Obviously any reduction in the number of times the HIE has to query HIS will improve the performance.
How to improve HIE performance and usability:
In the increasing order of implementation complexity:
- Browser Cache: As the sessionStorage for browsers mature, as much of the available information must be cached. For example, the matching records returned by patient search query could be cached so that if the opened record is not the correct one, other records can be accessed without issuing a duplicate search. Usage of localStorage might lead to privacy concerns.
- Server Cache: Caching at the server level using technologies such as Ehcache and HazelCast may also reduce HIS access and latency.
- As I mentioned in the phonegap article, a webapp is ideal for ease of maintenance as no mobile specific functions are utilized in clinical viewers. However, a native app may be good from a UX perspective.
- Hybrid HIE server: Hybrid HIE combines a centralized HIE and a federated HIE. In essence Hybrid HIE server permanently stores some of the information gathered from HIS systems to reduce latency.
- Design HIS systems from the ground up to be HIE aware.
I shall soon discuss the concept of HIE aware HIS along with architectural and implementation ideas. Clinicians rejected HIS from the billing world that did not fit their clinical workflow. A conceptual breakthrough is needed to save eHealth. Do you have any ideas?
- Kedro for multimodal machine learning in healthcare - January 25, 2023
- Using OpenFaaS containers in Kubeflow - December 14, 2022
- Six things data scientists in healthcare should know - November 3, 2021