Austin Health deploys AI to monitor COVID-19 symptoms

Medical centre rolls out COVID-19 monitoring system

Victoria-based Austin Health has deployed a series of pioneering AI-infused solutions that are helping to improve COVID-19 patient care. This will “empower the community, and reducing the administrative burden facing its hospitals”.

Austin Health’s community self-assessment platform – developed over the space of just seven days –attracted 25,000 people on day one of its release, with over 2000 completing the assessment. This  allowed the hospital to advise more than four out of five people taking the assessment that they did not need to present at hospital – helping to free up resources and reducing the potential exposure of people deemed at low risk .

It has also rolled out a COVID-19 symptom monitoring system for at-home patients reporting severe or deteriorating symptoms.  The solution includes leading-edge capabilities, such as an enhanced respiratory audio analytics feature and links the patient to a clinician when required.  The organisation has also deployed an appointment management system that has dramatically improved administrative efficiencies across the hospital.

Austin Health is a Melbourne-based hospital and one of the largest Victorian providers of training for specialist physicians and surgeons. In the last financial year, it completed 114,379 inpatient admissions, 89,675 emergency attendances, 28,902 surgical procedures and 279,156 specialist clinic appointments. Its researchers have been involved in 1,300 clinical trials.

The health  facility worked with Microsoft data and analytics partner Arden Street Labs, Austin Health has developed and deployed its series of intelligent, innovative solutions, called COVID-Care.

These include:

  • A digital platform to help emergency department staff collect data from people arriving at hospital;
  • A self-assessment tool allowing people to input their symptoms remotely and then be told whether or not they need to go to hospital;
  • A symptoms management solution that uses AI to assess COVID-19 patients’ respiratory and other targeted symptoms; and,
  • A secure portal allowing patients to access results and manage appointments and rescheduling.
  • The symptoms management tool asks patients to record a number of observations daily, including an advanced respiratory feature to analyse audio recordings using a number of AI/ML techniques, and leveraging Azure Cognitive Services, for any evidence of a shortness of breath.
  • If the system identifies a potential problem, then clinical personnel are alerted and able to provide further advice or support to the patient.
  • Two additional solutions are in the wings:
  • An in-clinic symptom assessment, helping to standardise risk of COVID for in-hospital patients to help alleviate the strain on the infectious disease team; and
  • An all of hospital visitor registration system, helping to digitally capture who is at the hospital, their risk of COVID infection, supported by QR code functionality and PowerApps for check-in.

Alan Pritchard director of EMR and ICT Services at Austin Health said the key to the success of the solution was the direct involvement of the hospitals’ clinical personnel who played an important role in the hospital’s agile approach to systems development.

The systems had been accelerated to help Austin Health cope with the COVID-19 crisis, these types of solutions were on the technology roadmap and some could provide enduring benefits, he said.

Austin Health also use machine learning to assess people’s respiratory symptoms.

“[This] applies to any person with a chronic respiratory condition, of which there are many conditions and millions of people. I think that particular solution, if it works well over the next few weeks, may have some enduring potential.”

Jeff Feldman CTO of Arden Street Labs said the key underpinning technologies used to develop and deploy the solutions include Azure Machine Learning, Azure Maps, Web Apps containers, DataBricks, PowerApps and Power BI.

“We are using Azure Databricks and Azure ML to support the key analytics capabilities built into the solutions, we’ve taken a very risk-adverse approach putting client confidentiality at the forefront,” he said. “Therefore, everything employed from Azure has been built leveraging Azure’s security features, including highly restricted access and compartmentalised approach to data.”

 

Tags:

Leave a Comment

Related posts