Hong Kong faces a crisis. The city has just 40 palliative medicine specialists for 7.5 million people. That’s one specialist for every 150,000 to 180,000 residents – far below global standards.
New legislation making advance medical directives legally binding in May will make this shortage even more acute. University of Hong Kong researchers think AI might help solve the problem.
They’re building a chatbot to train healthcare professionals in end-of-life communication at scale. The tool could help thousands of doctors learn these complex conversations quickly.
How does it work?
The chatbot simulates realistic patient conversations using validated assessment criteria. Students and doctors practice advance care planning discussions while receiving real-time feedback on their communication skills.
“This is a conversational AI simulator that enables students and clinicians to practice ACP discussions in realistic scenarios with real-time feedback,” says Dr Jacqueline Yuen Kwan-yuk, research lead at HKU Li Ka Shing Faculty of Medicine.
The system evaluates key behaviors like information sharing, understanding patient preferences, responding to emotions, and shared decision-making. It’s built on the Advance Care Planning Communication Assessment Tool (ACP-CAT), which researchers tested on 137 real conversations between doctors, patients and families.
The chatbot trains on real conversation transcripts. The team expects 24-36 months of development, testing first with students, then practicing clinicians.
Why does it matter?
Current training methods can’t scale fast enough. Traditional apprenticeship models can’t reach the thousands of clinicians who need upskilling before the new legislation takes effect.
Quality problems already exist in advance care planning conversations. In the HKUMed study, doctors discussed backup decision-makers in fewer than 5% of cases. They explored non-medical priorities in only 30% of conversations.
“Our AI-enabled training approach directly addresses these risks by rapidly building workforce capacity in a standardised, scalable, and evidence-based manner,” Dr Yuen says.
The AI platform offers 24/7 practice access without faculty supervision. It provides standardized training across institutions and personalized feedback at scale. Doctors get a safe space to rehearse emotionally complex conversations.
Poor training could create “superficial compliance” where medical directives are legally valid but don’t reflect patient wishes. It might also cause moral distress among unprepared providers and fuel public skepticism about the whole system.
The context
Hong Kong’s specialist shortage is severe. Global recommendations call for at least 1.5 palliative care specialists per 100,000 people. Hong Kong has 0.6 specialists per 100,000 residents.
The shortage of trained conversation facilitators is “even more acute,” according to Dr Yuen. Few structured programs exist to build these communication skills among practicing doctors.
Other regions are tackling similar challenges. Singapore launched myACP, a free online tool for documenting end-of-life care preferences. It’s available to citizens aged 21 and over without serious medical conditions.
Virtual reality is also being tested for training. Tokyo-based Jolly Good partnered with Harvard-affiliated Brigham and Women’s Hospital to create VR modules for palliative care communication.
The Hong Kong team plans to embed their assessment framework into routine clinical workflows. Healthcare institutions could use it for quality improvement audits and peer reviews. It might also integrate with electronic health records through structured documentation templates.
Dr Yuen emphasizes that AI should support, not replace, clinical judgment in these deeply personal conversations. Any system would need robust privacy safeguards and full compliance with Hong Kong’s data protection laws.