⚡ Quick Summary

Implementing artificial intelligence (AI) on a national scale in healthcare poses significant challenges. At the HIMSS24 APAC session titled “Implementing AI-Powered Triage for a National Virtual and Digital Front Door,” Dr. Nirvana Luckraj, Chief Medical Officer of Healthdirect, shared insights on how they successfully integrated an AI-driven clinical decision support system (CDSS) to enhance their national helpline services.

💡 Key Features and Benefits

• 🔑 24/7 Helpline Services: Healthdirect provides free access to various helpline services, including nurse triage and GP telehealth, catering to specific groups such as pregnant individuals, the elderly, and those facing mental health challenges.
• ⚙️ Integrated Health Information: The platform offers information on medications and curated health content, along with online booking and referrals that connect directly to national care pathways like My Health Record.
• 📊 Advanced Triage Capabilities: The new CDSS allows for the analysis of multiple symptoms and risk factors, improving the accuracy and comprehensiveness of patient assessments.

👩‍⚕️ Addressing Challenges

• Healthdirect’s transition to a new CDSS was driven by the need for a more sophisticated tool that could handle complex symptomatology.
• Dr. Luckraj emphasized the importance of a system that reflects real consultations, taking into account various risk factors such as chronic history and travel history.
• The previous CDSS relied on a decision tree algorithm, which was limited to single symptom inputs and required extensive maintenance.

🔍 Implementation Process

• Healthdirect evaluated five global vendors for an AI-based CDSS, conducting clinical tests to assess their offerings.
• Collaboration with AI experts from Macquarie University’s Australian Institute of Health Innovation helped validate the system’s accuracy through case vignettes.
• Ultimately, Healthdirect selected Infermedica’s AI CDSS, which utilizes a probabilistic Bayesian inference engine to enhance decision-making.

🚀 Overcoming Resistance

• Following the launch of the new CDSS in March 2023, staff initially resisted the change, finding it challenging to adapt to the new system that required inputting multiple symptoms.
• Feedback from nurses led to improvements, including a feedback loop mechanism allowing them to adjust triage outcomes based on their clinical judgment.
• Dr. Luckraj noted that the CDSS is designed as a support tool rather than a diagnostic system, enabling nurses to modify outputs as needed.

📈 Positive Outcomes

• One year post-implementation, nurses reported increased satisfaction with the new user interface and a higher agreement rate with the CDSS outputs.
• Healthdirect has not recorded any serious clinical incidents linked to the CDSS, and patient satisfaction scores have reached their highest levels in three years.
• The AI-powered CDSS has also been integrated into Healthdirect’s online symptom checker, providing users with access to various care options, including urgent care services.

🔗 Future Plans

• Healthdirect aims to enhance consumer experience by offering tailored information based on symptom inputs.
• Since the introduction of AI, the helpline has successfully diverted half of emergency calls to less acute services, connecting 44,000 calls to virtual emergency departments and booking nearly 5,500 online appointments for urgent care.
• Overall, Healthdirect has advised approximately 350,000 individuals on managing their symptoms at home, demonstrating the effectiveness of AI in reducing hospitalizations.

🔗 Sources

• Artificial Intelligence and Decision-Making in Healthcare | PMC
• AI in Medical Devices and Clinical Decision Support Systems | PMC
• AI-Based Clinical Decision Support Systems in Cardiovascular Diseases | PMC
• Accessing AI for Clinical Decision-Making | PMC
• Developing an AI-Assisted Clinical Decision Support System | PLOS ONE