Exploring Gen AI for Healthcare: What Could Ultrasound Look Like in a ChatGPT World?

A clinician and a patient view data on a tablet.

ChatGPT put a spotlight on generative artificial intelligence (gen AI). The platform set a record for the fastest-growing consumer user base; it surpassed one million active users five days after it launched and grew to an estimated 100 million users in only two months1. The interest and use of generative AI across business sectors has also steadily increased. According to a 2023 McKinsey Global Survey, about one-third of organizations surveyed are using generative AI tools regularly in at least one business function. This widespread adoption is accompanied by high expectations, with three-quarters of respondents anticipating significant or disruptive changes due to generative AI in their industries within the next three years2.

Although the healthcare industry is navigating the early days of its AI transformation, generative AI opens up a new realm of potential. Jan Beger, Head of AI Advocacy at GE HealthCare notes, “There are many opportunities to solve key critical challenges across hospitals and health systems with AI and specifically generative AI and foundation models. However, it is a process to uncover the right applications to make an impact, drive adoption, and build trust.”

Generative AI for Healthcare

Artificial intelligence is a wide-ranging field that encompasses several specialized areas. Machine learning (ML), which operates by learning patterns and rules from pre-existing data, has been applied in various fields for decades. ML is akin to a student learning from textbooks and applying those rules to solve problems. It excels in tasks such as image pattern recognition, outcome predictions based on historical data, and generating product recommendations. Deep learning (DL) is a subfield of ML that teaches computers to learn and make decisions by using a lot of data and mimicking how the human brain works. In the healthcare sector, applications of these capabilities can include analyzing medical images, assessing disease risks, and crafting treatment plans based on historical patient data. ML and DL based technologies are used in practice of ultrasound to automatically identify organs, tissue, and structures during a scan, assess the quality of images, and flag abnormalities in an image that should be investigated by a clinician.

Gen AI is a more recent and advanced form of artificial intelligence. It functions like a creative artist, not just following existing instructions but using learned data to create entirely new content, such as images or text. This type of AI has the capability to envision and simulate scenarios it has never directly experienced before. Generative AI examples in healthcare could include creating realistic medical images for training or simulation purposes or modelling the progression of diseases in virtual environments.

Copy of AI in Healthcare WIDE (2).svg

Generative AI may be particularly well-suited to healthcare, which is, unfortunately, known to grapple with vast amounts of unstructured data. Unlike ML and DL that rely on supervised learning with well-labeled data, generative AI can employ unsupervised learning techniques to both structured and unstructured data. This ability makes it invaluable for applications involving electronic medical records (EMRs) and physician’s notes, where it could help automate documentation, support clinical decision-making, enhance patient engagement, and even drive forward population health research by identifying trends and patterns within large data sets. In fact, one study found that clinical notes generated by ChatGPT were on par with those written by senior internal medicine residents, with the authors suggesting the technology might be ready for a larger role in everyday clinical practice3.

Potential Applications of Generative AI in Ultrasound

Ultrasound plays a crucial role in diagnosis and treatment. Its non-invasive nature and real-time visualization capabilities have made it indispensable in care settings such as obstetrics, cardiology, and emergency medicine, to name a few. However, the acquisition of ultrasound imaging is highly dependent on the skill set and experience of the sonographer or clinician conducting the exam. Additionally, the interpretation of ultrasound images can be subjective and heavily reliant on the expertise of the interpreting physician. Ultrasound imaging is also a field facing operational challenges, including a sonographer shortage, increasing demand, and the enterprise management of scans captured at the bedside.

Enter generative AI, which may have the potential to transform the landscape of ultrasound imaging. This advanced form of AI can handle the complexities associated with ultrasound data—ranging from image acquisition and processing to interpretation and training. There are opportunities to leverage it in several areas.

Enhancing Diagnostic Precision and Speed

Generative AI holds immense promise in improving scanning accuracy by guiding real-time ultrasound image acquisition, which may be particularly valuable in settings focused on developing the expertise of a less experienced ultrasound technologist team.

Ravi Nagavarapu, Senior Global Product Manager at GE HealthCare, underscores the transformative potential of AI in ultrasound imaging, emphasizing its role in enhancing accuracy and accessibility.  He said, "AI-powered image acquisition can empower sonographers and clinicians, leading to consistent results and improved patient care.” 

Quotes Gen AI (2).svg

Gen AI could also support image quality. Despite technological advancements, ultrasound images are sometimes susceptible to artifacts and noise. Gen AI algorithms may be able to intelligently enhance these images, removing noise, sharpening details, and improving overall image quality.

Additionally, gen AI models have the potential to analyze ultrasound images and automatically generate detailed reports that describe their findings. These reports could include measures, annotations, and preliminary assessments which would require additional manual input by a physician, but would allow them to move more efficiently through their workflow and reduce the reporting turnaround time.

Supporting Clinical Decision-Making and Efficiency

Systems that integrate gen AI have the potential to analyze extensive patient datasets to provide healthcare providers with critical insights, which could assist in complex decision-making processes. This could mean suggesting appropriate tests based on patient history and standardizing interpretations across various medical images, reducing variability and ensuring consistent care delivery across different healthcare facilities.

For physicians, gen AI could augment and assist by detecting abnormalities, and providing comprehensive training simulations. Additionally, the automation of routine and repeatable tasks by generative AI could lead to efficiency gains, optimizing the workflow and therefore allowing clinicians to focus more on patient care rather than administrative tasks or documentation.

Improving the Patient Experience

Given the complexity of healthcare information and clinical recommendations, there are potential gen AI use cases that could help patients better understand and interpret their conditions and recommended courses of treatment.  “Imagine a patient-facing version of a report that is generated from all of the physician and specialist recommendations that summarizes what a patient needs to know and understand to appropriately manage their care and treatment. This could be a great application of gen AI to empower patient engagement,” said Beger.

In addition to enhancing clinical workflows, generative AI can streamline operational processes and patient flows within healthcare institutions. By automating appointment scheduling and predicting equipment maintenance needs, AI-driven solutions could optimize resource utilization, reduce patient wait times, and minimize treatment delays.

Turning Gen AI Buzz into Next Steps

The integration of generative AI applications into ultrasound holds a lot of promise but the path there will require careful navigation. The healthcare industry stands at a pivotal juncture where collaboration, innovation, and ethical responsibility must converge to harness the full potential of AI technologies. It is imperative for stakeholders across the spectrum—from technology providers and healthcare institutions to policy makers and regulatory bodies—to engage in ongoing dialogue.

“As we appreciate these advancements, we’re also met with a crucial challenge: ensuring AI’s integration into healthcare is guided by a strong ethical compass, especially in sensitive areas like patient care and data management,” said Beger.

Quotes Gen AI (1).svg

GE HealthCare is focused on connecting the dots on AI and healthcare.  For the last three years it has topped the U.S. Food and Drug Administration (FDA) list of artificial intelligence enabled medical devices—it now has seventy-two 510(k) clearances or authorizations in the United States. Learn more about Verisound Digital and AI ultrasound solutions, which are simplifying the complex, eliminating non-value add tasks, integrating and automating where possible, and adding intelligence.

  1. Krystal Hu, “ChatGPT sets record for fastest-growing user base - analyst note,” Reuters, last modified February 2, 2023.  https://www.reuters.com/technology/chatgpt-sets-record-fastest-growing-user-base-analyst-note-2023-02-01/
  2. Michael Chui et al., “The state of AI in 2023: Generative AI’s breakout year,” McKinsey & Company, last modified August 1, 2023.  https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023-generative-AIs-breakout-year
  3. Ashwin Nayak et al., “Comparison of History of Present Illness Summaries Generated by a Chatbot and Senior Internal Medicine Residents,” JAMA Internal Medicine 183, no. 9 (2023): 1026–1027.