The race to apply large language models (LLMs) to healthcare is intensifying, with Anthropic announcing its new healthcare AI platform, Claude for Healthcare, on Sunday. This launch follows closely on the heels of OpenAI’s unveiling of ChatGPT Health, signaling a growing trend of tech companies entering the medical AI space. Both platforms aim to assist patients and providers, but Anthropic’s offering appears to focus more heavily on streamlining administrative tasks for healthcare professionals.
Anthropic’s new suite of tools is designed for use by providers, payers, and patients, and crucially, will allow users to connect their health data from various sources like smartphones and wearables. Like OpenAI, Anthropic has stated that this data will not be used for model training, addressing key privacy concerns. The initial rollout is expected to prioritize backend applications for healthcare organizations.
Anthropic’s Claude for Healthcare: A Deeper Dive
While ChatGPT Health is initially positioned as a direct-to-consumer chat tool, Claude for Healthcare is launching with a focus on “agent skills” and integrations with existing healthcare databases. These “connectors” provide Claude with access to resources like the Centers for Medicare and Medicaid Services (CMS) Coverage Database, the International Classification of Diseases, 10th Revision (ICD-10), the National Provider Identifier Standard, and PubMed. This access is intended to improve the accuracy and efficiency of the AI’s responses.
Streamlining Prior Authorization
One key application highlighted by Anthropic is the automation of prior authorization reviews. This process, where healthcare providers seek approval from insurance companies for certain treatments or medications, is often cited as a significant administrative burden. According to Anthropic CPO Mike Krieger, clinicians spend a disproportionate amount of time on paperwork rather than direct patient care.
Claude for Healthcare aims to alleviate this by automatically completing and submitting the necessary documentation, potentially accelerating approval times and freeing up clinicians’ schedules. This focus on administrative efficiency represents a strategic difference from the initial emphasis of OpenAI’s ChatGPT Health. The company believes automating these tasks is a more appropriate use of LLMs than providing direct medical advice.
Addressing Concerns About LLM Accuracy
The introduction of these platforms isn’t without its critics. Industry professionals have expressed concerns about the potential for “hallucinations” – instances where LLMs generate incorrect or misleading information – when applied to sensitive medical contexts. The accuracy of information is paramount in medical AI, and incorrect guidance could have serious consequences.
However, Anthropic’s approach of grounding Claude in established databases and focusing on tasks like prior authorization may mitigate some of these risks. By relying on verified information sources, the AI is less likely to generate fabricated responses. Both Anthropic and OpenAI consistently advise users to consult with qualified healthcare professionals for reliable and personalized medical guidance.
The Growing Landscape of AI in Healthcare
The launch of both Claude for Healthcare and ChatGPT Health underscores the increasing interest in leveraging artificial intelligence to improve healthcare delivery. According to OpenAI, approximately 230 million people are already discussing their health with ChatGPT each week, demonstrating a clear demand for AI-powered health information. This widespread adoption, even in its early stages, is driving further investment and development in the field.
The potential benefits of AI in healthcare are substantial, ranging from improved diagnostic accuracy and personalized treatment plans to reduced administrative costs and increased access to care. However, realizing these benefits requires careful consideration of ethical implications, data privacy, and the need for ongoing validation and refinement of AI models. The integration of LLMs into existing workflows also presents challenges, requiring training and adaptation for healthcare professionals.
The development of these tools also highlights the broader trend of applying LLMs to specialized domains. While general-purpose LLMs like ChatGPT have demonstrated impressive capabilities, tailoring them to specific industries, such as healthcare, can unlock even greater value. This involves not only training the models on relevant data but also developing specialized connectors and workflows that address the unique needs of the sector. The use of secondary keywords like “digital health” and “patient engagement” are also becoming more prevalent in discussions around these technologies.
The next steps for both Anthropic and OpenAI will involve gathering user feedback, refining their models, and expanding the functionality of their platforms. It remains to be seen how quickly these tools will be adopted by healthcare organizations and patients, and how effectively they will address the challenges of accuracy, privacy, and integration. Ongoing monitoring of regulatory developments and industry best practices will also be crucial as the field of medical AI continues to evolve.
