Applications of AI in the medical devices sector

Tsinghua University in China is developing a so-called “Agent Hospital”: a complete digital simulation of the medical treatment process, from disease outset to follow-ups post-treatment. LLMs support the doctors, nurses, and patients so they can perform tasks and communicate autonomously in this virtual environment.  

The concept involves treating virtual patients with AI doctors. Tsinghua University uses large medical information databases (e.g., DNA sequencing, personal health, hospital, and patient records) and ML technologies to train their AI systems, making them capable of reasoning and medical diagnosis. As of 2024, the “hospital” has 14 AI doctors and four AI nurses.  

Initially, the Agent Hospital aims to enhance its production process and use the AI hospital to train medical students to develop treatment plans. By simulating AI patients, students can practice developing treatment plans without harming real patients.  

Virtual doctors are scheduled to be implemented into patient care at hospitals across China. The team at Tsinghua University claims that 10,000 patients could be treated in a few days, whereas human doctors would require at least two years to treat that number. The successful introduction of the Agent Hospital would address challenges in healthcare services by enhancing patient outcomes, healthcare operations, training, and access to high-quality, affordable healthcare. 

However, AI doctors can only aid and will never fully replace humans, as uncertainty in medical practice is common. For example, the same disease can have different characteristics depending on the individual patient, and so treatment diagnoses and treatment options will vary.

In April 2024, GE Healthcare introduced Caption AI, software designed to facilitate quick and accurate cardiac assessments at the point of care when using their Vscan Air SL wireless handheld ultrasound device. The Vscan Air SL with Caption AI debuted at the 2024 American College of Cardiology Annual Scientific Session & Expo in Atlanta in April 2024.  

This software aims to help healthcare professionals efficiently capture cardiac images during routine checkups, aiding in early disease detection and improving patient outcomes. The Vscan Air SL with Caption AI ultimately helps simplify the process of capturing cardiac images, which means those with limited ultrasound expertise can also conduct rapid cardiac assessments effectively. The Vscan Air SL device itself is noted for its portability and imaging capabilities, featuring single-crystal transducer technology that delivers high levels of resolution, depth, and sensitivity.  

Caption AI offers real-time visual guidance to prompt users on probe movements and incorporates a quality meter to maximize the quality of the image. Upon image capture, the Auto Ejection Fraction (AutoEF) function swiftly calculates the left ventricular ejection fraction (LVEF), enhancing diagnostic efficacy. Moreover, users benefit from efficient scanning through features like AutoCapture, which automatically saves clips. Another feature is Save Best Clip, where you can select the best image loop for each view. 

Cardiovascular disease (CVD) is the primary cause of global mortality, and its prevalence is projected to increase with population growth and aging. Timely detection of heart disease is crucial for better patient outcomes, yet providing swift echocardiographic assessments at the point of care poses challenges for resource-limited facilities and practices. The Vscan Air SL with Caption AI addresses this industry pain point by simplifying cardiac image capture, making it accessible even to non-expert ultrasound users.  

This innovation lowers the barrier for healthcare professionals to conduct rapid assessments, especially in settings where skilled sonographers are scarce.