West Virginia University Develops AI to Transform Heart Disease Diagnosis

Researchers at West Virginia University are advancing artificial intelligence (AI) technology to enhance heart disease diagnosis in rural areas. This initiative addresses a significant gap in healthcare, as traditional AI models largely reflect data from urban populations, often neglecting the unique needs of rural patients. The project aims to create more accurate diagnostic tools tailored specifically for these underserved communities.

Prashnna Gyawali, an assistant professor in the Benjamin M. Statler College of Engineering and Mineral Resources, leads the effort within the Lane Department of Computer Science and Electrical Engineering. He notes that while AI is being increasingly utilized in healthcare worldwide, it predominantly relies on urban datasets. This reliance can lead to biases that compromise the effectiveness of AI in rural settings. “You have to ensure your algorithms have seen the populations where you want them applied,” Gyawali stated.

To tackle this issue, Gyawali and his research team are exclusively using data from rural patients in West Virginia. They have compiled anonymous datasets from various regions and are currently testing different AI models to evaluate their performance in diagnosing heart disease based on patient test results. This approach aims to ensure that the algorithms reflect the biological and socio-economic characteristics of rural populations.

The implications of these AI models are potentially transformative for rural healthcare. Gyawali emphasized the significant benefits they could provide, particularly in alleviating the burden on healthcare professionals who often face high demand with limited resources. “Health care problems are growing and we have manpower shortages,” he explained. In many cases, patients may need to travel several hours to receive a proper diagnosis, which can delay necessary treatment.

By introducing AI-powered diagnostic tools at local clinics, Gyawali envisions a future where early detection of heart disease becomes more accessible. “If we have more clinics with inexpensive scanning devices with an AI system attached, we can have an early detection system flagging certain patients,” he said. The goal is to create a reliable system that accurately identifies those in need of immediate care.

While the initial testing of AI models is promising, Gyawali emphasizes the importance of refining these technologies before they can be deployed in real-world settings. The current models have interacted only with historical rural datasets, and further validation is crucial. “Whenever we talk about safety-critical applications like health care, we need to make sure they’re reliable,” he stated, highlighting the risk of misdiagnosis that could arise from faulty AI predictions.

The team is committed to enhancing the AI model’s performance through ongoing research. Questions remain about how to further validate these algorithms and whether they can be effectively tested in clinics not involved in the initial study. Gyawali also expressed interest in extending the research beyond West Virginia to see if the model can be adapted for use in other states.

Looking ahead, Gyawali recognizes the necessity of policy-level interventions to facilitate the integration of AI tools into clinical settings. “That’s the roadmap toward adopting these tools in clinics,” he concluded. While there is no set timeline for when clinical trials may commence, the research team is dedicated to ensuring the technology meets the highest standards of safety and efficacy before it can be utilized in patient care.