Leveraging GenAI for High-Quality Clinical Support
Generative AI, particularly large language models (LLMs), are revolutionizing various sectors, with healthcare standing out as a field ripe for innovation. Since the launch of ChatGPT by OpenAI in November 2022, LLMs have showcased their potential to simulate human conversation, create new content, and interpret complex data sets. This month’s newsletter explores the applications of generative AI in healthcare, particularly in providing high-quality clinical support, and offers insights into the pathway for successful implementation.
"In healthcare, AI is not about replacing doctors, but about augmenting them. AI can help doctors do their jobs better, faster, and more efficiently." — Andrew NG, AI researcher and entrepreneur
The Role of Generative AI in Healthcare
Generative AI models are trained on vast amounts of data to generate human-like text. Other notable LLMs include Google’s Gemini and tools like Microsoft Copilot . These models have demonstrated capabilities beyond simple text generation, including passing medical exams, writing research articles, and interpreting electronic medical records (EMRs). This is particularly important in healthcare, where these models can enhance clinical support by improving predictive accuracy, simplifying model development, and reducing deployment costs.
Current State and Challenges
Despite the excitement, the implementation of generative AI in healthcare faces several challenges. A recent article from NPJ Digital Medicine highlighted that current AI models often lack generalizability and face data privacy issues. Many models are trained on limited datasets, which restricts their applicability across different healthcare systems. Additionally, concerns about AI hallucinations or generating incorrect responses when data is insufficient pose significant risks in clinical settings.
"More than any prior technological advancement in healthcare, the unique power of genAI requires that we temper our enthusiasm with both caution and skepticism." —?Dr. Haveh Shojania, Vice Chair for Quality & Innovation in the University of Toronto's Dept. of Medicine
Six Criteria to Evaluate Generative AI Models
To address these challenges, NPJ Digital Medicine proposes the use of an evaluation framework based on six criteria:
This framework aims to help healthcare systems assess the clinical value of AI models comprehensively. For instance, models integrated into EMRs can be evaluated on their ability to accurately predict patient outcomes and streamline clinical workflows.
Practical Implementations and Innovations?
Recent partnerships and developments underscore the potential of generative AI in healthcare. For example, Microsoft’s collaboration with Epic integrates generative AI to automate responses to patient messages, potentially saving significant time for healthcare providers. Similarly, Oracle Cerner's Oracle Clinical Digital Assistant utilizes multimodal AI to assist in notetaking, order medications, and manage appointments, enhancing both provider and patient interactions.
Accelerating Software Development with ChatGPT
A study conducted by a research team from NYU Langone Health highlights how generative AI can accelerate the development of digital health software, particularly in the context of clinical support. The study focused on leveraging ChatGPT to develop personalized automatic messaging systems (PAMS) aimed at enhancing clinical support. By harnessing ChatGPT’s capabilities, the researchers aimed to streamline the coding process and accelerate software development for various clinical support applications.
The findings revealed promising results, demonstrating the significant impact of ChatGPT on the software development lifecycle. Moreover, ChatGPT proved to be a powerful tool for bridging the communication gap between technical and non-technical team members, leading to a substantial reduction in development time for clinical support software.
Leadership, Incentives, and Regulations?
For generative AI to transcend being a health-tech fad, strategic leadership, appropriate incentives, and stringent regulation are essential. Leadership must focus on guiding AI development, ensuring model validation, and clarifying implementation guidelines. A regulatory framework, potentially led by bodies like the FDA, should address liability, data privacy, and bias within AI models. Incentives, particularly financial ones, will also be crucial to encourage widespread adoption. Generative AI tools should be viewed as capital expenses similar to EHR systems, necessitating both public and private investment to drive forward this technological advancement.
The Pathway Forward?
The adoption of generative AI in healthcare is at a pivotal moment. By leveraging the evaluation framework proposed by NPJ Digital Medicine , healthcare providers can better discern the true clinical value of AI models. However, a broader pathway involving defined leadership, regulatory clarity, and adequate incentives is critical to foster innovation and ensure responsible AI integration.
As generative AI continues to evolve, it promises transformative capabilities in clinical support, diagnosis, patient engagement, and administrative efficiency. The potential benefits are vast, but careful, strategic implementation is necessary to overcome the challenges and fully realize the technology’s potential.
Let's stay connected! Subscribe to the NantHealth Blog .