AI, cloud helping tackle sepsis’ deadly toll
- 20 November, 2019 06:00
Half a dozen New South Wales public hospitals are part of a point-of-care testing (POCT) trial that uses Microsoft’s Azure services in conjunction with mobile devices to help deliver real-time diagnostic test results to clinicians.
Buladelah, John Hunter Children’s, Orange, Trangie, Wagga Wagga and Westmead hospitals are participating in the trial, which involves the use of secure 4G LTE connections as well as NSW Health internal networks to upload data to Azure.
POCT can deliver pathology results outside of a laboratory, allowing health services to more quickly treat patients.
NSW Health Pathology offers POC testing for a range of conditions including cardiac damage, kidney health, diabetes and infection, with the organisation responsible for the world’s largest managed POCT service. It currently has more than 680 devices in use at more than 200 hospitals across the state.
“Point of care testing is vital to providing immediate, accurate results for certain pathology tests which can save lives in locations that don’t have 24/7 labs, so by enabling more services to offer this testing we’re helping clinicians diagnose and treat more patients more quickly,” NSW Health Pathology’s chief information officer, James Patterson, said in a statement.
Azure is also helping underpin the Proactive Sepsis Management project at Blacktown and Westmead hospitals. The world-first trial leverages Azure’s AI services and combines vital signs data, including from devices not directly integrated into clinical systems, with blood gas POCT data to help identify patients that are at risk of sepsis in the hospitals’ emergency departments.
Sepsis is a potentially deadly organ dysfunction resulting from an infection. According to the Australian Sepsis Network, organ failure and death are more likely if the condition is not recognised early and swiftly treated. In Australia it kills around 5000 people every year, ASN says.
NSW Health Pathology has worked with Microsoft on the trial, which uses a range of the company’s cloud technologies to analyse data from the Raspberry Pi-based POCT devices as well as other sources and combines it with patient demographic data to predict the risk of sepsis. The approach avoids the need to manually transpose data between different systems.
CosmosDB, IoT Hub, Event Hub, DataBricks, Power BI and Dynamics 365 are all being employed for the trial.
“Being able to run algorithms on patient information as it’s transmitted to their electronic medical record and apply analytics to help doctors detect when a patient is at risk of this condition is highly promising,” Patterson said.
“Early detection and therefore earlier treatment will help save lives and help reduce the number of people suffering long-term complications.”