Practice nurses screening for heart failure in Australian primary care
PANACEA-HF trains practice nurses to deliver AI-mediated cardiac ultrasound in metropolitan and rural-remote primary care clinics across Australia. Early data shows imaging quality rising sharply with structured training, and a clinical algorithm that outperforms unaided GP referral.
At a glance
Heart failure caught only at the hospital door
Heart failure most often goes undetected until a patient is acutely hospitalised, a costly and avoidable outcome. In primary care, even among high-risk individuals, the syndrome remains chronically missed. The conditions that cause it develop silently in the community over years, frequently in people already known to their GP.
The challenge has never been identifying who is at risk; it has been having a reliable, practical way to act on that risk before crisis strikes. PANACEA-HF was designed to test whether a structured, AI-supported screening pathway run by practice nurses can close that gap.
Practice nurses, AI-PoCUS, and a clinical algorithm
PANACEA-HF (Practice Nurses to Augment the Clinical Evaluation and cAre of people at high-risk of Heart Failure) is a prospective, pragmatic, multi-centre surveillance study and nested RCT conducted across metropolitan and rural-remote primary care clinics in Australia. The program is led by Professor Simon Stewart at the University of Notre Dame Australia and funded by a $2 million Medical Research Future Fund (MRFF) grant.
The pathway screens patients aged 60 and over with common heart failure antecedents. Practice nurses with no prior imaging experience were trained over six months to apply AI-mediated portable cardiac ultrasound (AI-PoCUS), acquiring parasternal long-axis and apical chamber views with colour Doppler and tissue Doppler velocities.
The clinical algorithm
The algorithm integrates ESC-defined signs and symptoms, 12-lead ECG findings, NT-proBNP levels, and AI-PoCUS performed by the practice nurse. It produces a systematic pathway from initial profiling through to specialist referral or active surveillance, designed to catch heart failure earlier without flooding cardiology clinics with unnecessary referrals.
AI in the hands of non-specialist operators
Us2.ai is the AI software solution for the AI-PoCUS step of the pathway. The platform performs fully automated analysis on echocardiographic images regardless of who acquires them, producing the structured cardiac assessment that feeds into the algorithm. Because the software handles measurement and reporting end-to-end, the program does not require an on-site echocardiographer in every clinic.
That makes guideline-directed diagnostic quality achievable in primary-care and rural-remote settings that could not otherwise host a hospital-grade echo service. Practice nurses acquire the study with a portable probe and the AI generates the report used to triage the patient.
Imaging quality rises sharply with structured training
From the first 300 formally screened cases (150 men, 150 women, mean age approximately 71), a clear learning curve emerged and with it a clear threshold of competence. Imaging failure rates dropped from 16% in the first 50 cases to zero in the final 50, while the rate of complete AI-PoCUS reports generating more than 30 cardiac indices rose from 26% to 68%. Full report generation was 6.30-fold more likely by cases 251 to 300 versus the first 50 (p<0.001). Imaging success was unaffected by patient age, sex, or BMI; the gains were driven entirely by nurse experience.
A separate analysis evaluated the clinical algorithm against unaided GP clinical judgement. On an initial cohort of 100 patients with hypertension and diabetes, GP-led referral on clinical grounds achieved 100% sensitivity but only 88.8% specificity, generating nine unnecessary specialist referrals. The algorithm achieved 96.7% sensitivity and 98.8% specificity with just one false-positive referral, and a positive likelihood ratio of 77.3 (95% CI 11.0 to 543).
Full results from more than 700 screened patients across metropolitan, rural, and regional GP clinics in Australia are expected in August 2026.
A model for nationwide adoption
These findings demonstrate that practice nurses, with structured training and ongoing support, can reliably generate clinically meaningful cardiac imaging as part of a proactive heart failure surveillance program. For rural and regional settings where specialist resources are scarce, this kind of streamlined triage has particular value: it brings guideline-quality screening into the GP clinic rather than waiting for the patient to reach a cardiologist.
If the full results confirm the early pattern, the team plans to present the findings to the Australian federal government with the goal of nationwide adoption in every general practice.
Publications & announcements
PANACEA-HF: Practice nurse-led AI-PoCUS detects undetected heart failure in primary care
Heart Failure Congress 2026
Conference abstract · 2026PANACEA-HF: Clinical algorithm optimizes detection of undiagnosed heart failure in primary care
Heart Failure Congress 2026
Announcement · Jun 2025Nurses use AI Echo for Heart Failure detection in Australia
Us2.ai partnership announcement