Rural Appalachia faces a well-known challenge: soaring rates of heart disease paired with limited access to advanced medical tools for rural health heart care. Now, West Virginia University researchers are bringing innovation closer to home—literally. They’re using artificial intelligence to harness low-tech electrocardiograms (ECGs) and help rural clinicians identify heart failure before it’s too late.
The Rural Diagnostic Gap
Heart failure—the chronic condition where the heart struggles to pump enough oxygen-rich blood—is a serious health concern across the U.S. But it’s particularly relentless in rural Appalachian areas, where limited access to advanced diagnostics often means delayed detection.
Most AI models rely on data from urban hospitals, meaning they may not perform well for rural patients. A West Virginia woman working long hours, breathing in coal dust, and with minimal preventive care may be misdiagnosed—unless the AI understands her reality.
Why ECGs, and Why Now?
Echocardiography—the gold standard for measuring heart function—is costly and often inaccessible in rural clinics. In contrast, ECGs are simple, inexpensive, and widely available. WVU researchers tested whether AI could use ECG data to estimate a patient’s “ejection fraction,” a key metric for heart failure.
How the AI Was Built
Using anonymized records from more than 55,000 patients across 28 hospitals in West Virginia, the team trained several AI models. They compared deep-learning (neural networks) and non-deep-learning approaches. A model known as ResNet emerged as the most accurate, especially when tuned with data from specific ECG leads.
The Promise of Contextualized AI
By tailoring these models to rural data, the goal is to give local clinicians a powerful, timely tool—bridging the gap between advanced diagnostics and what’s available on-site. With heart failure affecting over 6 million Americans and one in four likely to experience it in their lifetime—or even higher rates in Appalachia—this innovation could make a life-saving difference.
Looking Ahead
While these AI models are not yet in clinical use—further validation is needed—the groundwork is in place. The researchers are already identifying which ECG lead combinations boost accuracy and planning to scale training with even more data. As models improve, they could become trusted partners for rural doctors.
Imagine a world where a simple, low-cost ECG at your local clinic not only captures your heartbeat—but, leveraging AI trained on patients just like you, helps catch heart failure early and equally. WVU’s research is turning that world into an emerging reality.
References
- WVU-trained AI models diagnose heart failure using ECGs in rural Appalachia (WVU Today / MedicalXpress; August 28–31, 2025)
- Intermountain’s coverage: “WVU scientists develop AI models that can identify signs of heart failure” (published Sep 2, 2025) .
