AI agents

What are the most common use cases of AI agents in healthcare? 

You get an agent and YOU GET AN AGENT! Everybody gets an agent!

AI agents are everywhere - or at least that what it sounds like in our world. But what are they really and what exactly do we hope AI agents will do that chatbots could never? Word on the street is that:

The major use cases of AI agents in healthcare are expected to enhance efficiency and patient experience by providing 24/7 support, automating appointment scheduling and reminders, assisting with billing inquiries, and streamlining workflows.

These applications are expected to hit the familiar healthcare KPIs of reduced wait times, optimized operations, and improved overall patient care.

What are the public health implications of AI agents spreading misinformation? 

AI-powered chatbots like ChatGPT have the potential to democratize access to health information but also pose risks by spreading misinformation, which we've talked about before. Research has shown that due to their inability to fully verify facts, AI agents can unintentionally spread false or biased health claims, which may disproportionately affect vulnerable communities. This is most dangerous when it comes to race-based practices in medicine that can cause generational impacts.

More specifically, experts have already warned that AI agents spreading misinformation pose serious public health risks by fueling vaccine hesitancy, climate change denial, and online misogyny. When it comes to public health implications, the threat is when AI agents amplify false narratives, erode trust in institutions, and contribute to disease outbreaks, such as the measles resurgence - where do people turn?

What are the infrastructural needs to run AI agents? 

Paradoxically, AI agents at scale do come with some hurdles. If everyone were to get an AI agent, there would be a massive strain on our infrastructure.

Running AI agents requires robust infrastructure, including secure ability and access, efficient reasoning and planning, and seamless component orchestration.

And when it comes to databases - context management is crucial, encompassing primary context (agent goals), direct context (real-time system states), and external context (general knowledge). But let's not get too nerdy about the details.

Simply put, running AI agents requires low-latency communication, high model accuracy, and resource constraints.  We need more conversation about what this looks like at scale and in every day life.

Who is building AI agents for healthcare and who is investing in them?

Still, several companies and investors are betting on AI agents in healthcare to enhance diagnostics, patient care, and efficiency. There are notable cases to follow: PathAI and Qure AI use AI for disease detection, while Google’s MedPaLM and Viz.ai improve medical imaging and decision-making. Nabla, Suki, and Dax Copilot automate administrative tasks, reducing clinician workload. Startups like Assort Health and Parakeet Health optimize patient interactions, and Hippocratic AI, backed by $141 million, launched a Healthcare AI Agent App Store. Somalab and Quadrivia AI focus on medical training and workforce efficiency. 

By next week, I'm sure there will be more...