Mayo Clinic cardiology leader Patricia A. Pellikka, MD’s optimistic vision of echo AI.

Artificial Intelligence

Automating Stress Echo

A new JACC study showed that Ultromics’ EchoGo Pro AI solution can accurately classify stress echocardiograms, while improving clinician performance with a particularly challenging and operator-dependent exam. 

The researchers used EchoGo Pro to independently analyze 154 stress echo studies, leveraging the solution’s 31 image features to identify patients with severe coronary artery disease with a 0.927 AUC (84.4% sensitivity; 92.7% specificity). 

EchoGo Pro maintained similar performance with a version of the test dataset that excluded the 38 patients with known coronary artery disease or resting wall motion abnormalities (90.5% sensitivity; 88.4% specificity).

The researchers then had four physicians with different levels of stress echo experience analyze the same 154 studies with and without AI support, finding that the EchoGo Pro reports:

• Improved the readers’ average AUC – 0.877 vs. 0.931
• Increased their mean sensitivity – 85% vs. 95%
• Didn’t hurt their specificity – 83.6% vs. 85%
• Increased their number of confident reads – 440 vs. 483
• Reduced their number of non-confident reads – 152 vs. 109
• Improved their diagnostic agreement rates – 0.68-0.79 vs. 0.83-0.97

The Takeaway

Ultromics’ stress echo reports improved the physicians’ interpretation accuracy, confidence, and reproducibility, without increasing false positives. That list of improvements satisfies most of the requirements clinicians have for AI (in addition to speed/efficiency), and it represents another solid example of echo AI’s real-world potential

The Wire

• CAC’s Statin Guidance: A new JACC study suggests that CT-based coronary artery calcium (CAC) scores are more effective for guiding statin therapy decisions than risk-based PCE scores (the current standard). The researchers provided physicians with either CAC or PCE scores to assess 540 patients, finding that the physicians were more likely to apply CAC scores to their statin recommendations than PCE scores (88% vs. 75%) and patients in the CAC group were far more likely follow their doctors’ orders after three months (62.2% vs. 42.2%). By the one-year mark, patients in the CAC group had superior statin adherence, lower LDL-C levels, and similar or lower healthcare costs.

• YNHH Home Hospital: Yale New Haven Health is partnering with Medically Home to launch a Home Hospital Program for local Medicare patients. Patients who would otherwise need to be hospitalized will instead receive a daily telehealth visit with a physician through a provided tablet, twice-daily in-person visits from a nurse, plus additional services. Although this isn’t really a cardiology story, patient care’s continued shift beyond hospital walls definitely has cardiology implications.

• Ultrasound-Embedded AI: A team of UK-based researchers detailed an AI algorithm that can distinguish normal and clogged carotid arteries using grayscale static duplex ultrasound images (DUS), potentially allowing less-experienced clinicians to perform these exams. Tested on 156 images, the algorithm detected normal scans with 91% sensitivity, 86% specificity, and 92% accuracy, and detected stenotic arteries with 87% sensitivity, 82% specificity, and 90% accuracy. 

• Healthcare Hiring Boost: The US healthcare sector recently saw a bump in hiring according to the May Jobs Report, adding 28k jobs throughout last month, including 16k hospital workers. Following hospitals, the subsectors with the largest workforce gains were ambulatory services, physician offices, and nursing care facilities, each adding ~6k employees. Despite the improvement, the healthcare workforce remains 1.3% (223k jobs) below pre-pandemic levels, compared to a 0.5% decline across all industries.

• K Health + Mayo Clinic: Virtual care company K Health teamed up with Mayo Clinic to launch an algorithm that helps clinicians on its platform deliver personalized hypertension treatment using insights from Mayo Clinic Platform Discover’s de-identified patient data. K Health’s hypertension treatment tool combines clinical data with outcome data to predict the medication and lifestyle changes most likely to help individual patients, and the language used in the press release makes it look like we could see the Mayo Clinic partnership expand to other conditions sooner rather than later.

• MRI CVD Markers: Analysis from the multigenerational Framingham Heart Study revealed that high thoracic aortic wall area (AWA) and arterial plaque volumes could serve as MRI-based biomarkers for future cardiovascular events. Researchers analyzed baseline MRIs from 1,513 middle-aged adults, finding that higher AWA and plaque were correlated with future CVD events (~13-year hazard ratios: 1.20 & 1.63), with elevated AWA most closely linked to stroke (HR: 1.32) and plaque volume to heart disease (HR: 2.20).

• Surgeon General Takes on Burnout: The U.S. Surgeon General issued a call to collectively address rising healthcare worker burnout and resignations. Dr. Murthy’s burnout recommendations include: (1) involving workers in process and culture improvement initiatives; (2) eliminating punitive policies for seeking mental health care; (3) ensuring sufficient compensation, labor levels, and safety; (4) reducing administrative burdens so workers can focus on patients; (5) prioritizing social connection & community as healthcare system core values; and (6) increasing public health focus on social determinants of health.

• Forgoing Treatment: Americans above the age of 50 are struggling with healthcare costs, with 40% concerned about their ability to afford care and others forgoing treatment or skipping medications altogether. That’s from a recent West Health and Gallup survey of 6.7k US adults, which found that nearly 25% of seniors have cut back on at least one basic need to afford healthcare. Healthcare’s financial burden was the heaviest for respondents aged 50-64, who have more health problems than younger groups but are still too young to qualify for Medicare, with 26% of this age bracket recently skipping treatment because of cost (vs. 12% of those 65+).

• Cardiac MRI AI Measurements: UK researchers developed and evaluated a cardiac MRI AI model, finding that the model’s automated CMRI measurements correlated better than manual measurements for left ventricular stroke volume (r = 0.74 vs. 0.68), pulmonary vascular resistance (r = 0.62 vs. 0.41), and pulmonary artery pressure (r = 0.56 vs. 0.37) with 178 patients. Moreover, AI-measured right ventricular end-systolic volume, ejection fraction, and mass all predicted mortality in 920 patients with pulmonary arterial hypertension (hazard ratios: 1.40, 0.76, and 1.15) over a 3.8yr average follow-up period.

• Inpatient Volume Projections: Vizient subsidiary Sg2’s 2022 Impact of Change Forecast predicts that inpatient volumes will quickly recover to pre-pandemic levels then grow only 2% over the next decade, but an increase in chronic conditions is expected to cause an 8% rise in inpatient days during the same period. Outpatient volumes are projected to grow 16% through 2032 (3% above population growth estimates), while home visits are forecasted at 19% growth, potentially useful indicators for anyone rethinking service line prioritizations or home care initiatives.

• Chronic Disease Funding: The Patient-Centered Outcomes Research Institute is offering up to $50M for studies demonstrating ways that telemedicine can help patients with chronic diseases. The nonprofit institute will fund research into interventions that incorporate telemedicine into primary care for the management of comorbidities, with an online application system that opens on September 7.