Prädiktives maschinelles Lernen entwickeln, mit Flink | Workshop am 18. Dezember | Jetzt registrieren
Telemedicine Applications
Confluent enables telemedicine providers to deliver a reliable service by streaming, processing, and governing healthcare data in a secure way.
Provide Reliable Remote Healthcare in Real Time
Telemedicine applications rely on the continual flow of data from multiple sources, including Electronic Healthcare Records (EHRs), diagnostic centers, and increasingly, remote healthcare trackers.
In order to deliver a reliable service which is responsive to the immediate needs of patients, telemedicine providers must be able to integrate and process healthcare data at scale and in real time.
With Confluent as the backbone of their streaming data architectures, telemedicine providers are able to:
Improve service reliability for patients, by ensuring that applications remain running in the event of data center outages or during periods of incredibly high demand.
Improve the experience of healthcare professionals, by facilitating the flow of data across multiple healthcare systems.
Manage security and compliance requirements, by encrypting personal information, preventing unauthorized access to data, and enforcing data retention policies.
Build with Confluent
This use case leverages the following building blocks in Confluent Cloud.
Streaming Architecture
Data Architecture
Streaming Architecture
Events are sent from clients to the Application Gatway, which serves both as API endpoint and as a message translator (into FHIR) for incompatible messages. Events are then sent to Confluent Cloud.
Flink is used to join and aggregate messages, and create new data products which form the basis of a sepsis altering system.
An external alerting application subscribes to the created stream, and provides sepsis alerts in real time based on the created topic.
Data Architecture
In this simplified example, a join is used to combine relevant data from four separate topics in order to create a new topic, ‘sepsis_monitoring’. A windowed aggregate function is then used to calculate a sepsis risk score per patient, and create a new topic, ‘sepsis_alerts’, which only contains patients with scores above a certain threshold. This topic is subscribed to by an external application which notifies health professionals in real time of patients at risk of sepsis.
- sepsis_monitoring – A topic consisting of medical data points indicative of patients’ current sepsis risk.
- sepsis_alerts - A topic containing information relating to patients who have surpassed a sepsis risk threshold. This topic can be subscribed to by multiple applications in order to streamline emergency responses, and can be stored and replayed for historical analysis (i.e., to measure the effectiveness of sepsis responses).