Skip to main content
Rural Health Information Hub
Rural Health Information Hub

AI Conversational Agents

Conversational AI, also known as conversational agents, simulates human interactions by using natural language processing and machine learning to support clinicians. Conversational agents are also called chatbots, conversational bots, virtual agents, virtual bots, virtual interactive agents, digital assistants, and smartbots. Unlike traditional chatbots that rely on pre-set scripts, AI-augmented conversational agents can draw on a person's information to personalize responses.

AI conversational agents have been used in different contexts and vary in the tasks they support and how well they work. Examples include:

  • Supporting healthcare staff with appointment scheduling, reminders, and prescription refills
  • Collecting patient history data and supporting disease diagnosis and monitoring
  • Providing patient education for people making lifestyle changes, and improving patient engagement in their care plans
  • Supporting therapy and counseling for mental health conditions.

Evidence on the effectiveness of conversational agents is emerging and mixed, with some evidence that they can support patients in smoking cessation, healthy habit formation (for example, physical activity, managing weight or sleep), and, when used in combination with therapy, reduce anxiety levels and improve psychological well-being.

AI conversational agents have potential to expand patient access, bridge resource gaps, and support clinicians in rural areas. These tools also have the potential to help socially isolated people by providing companionship. Conversational agents may naturally pair with clinical decision support functions like predictive analytics and remote patient monitoring services.

Examples of Rural Programs Using AI Conversational Agents

  • A mental health chatbot developed at Stanford University has been used to support adolescents and young adults in rural and underserved areas. The chatbot was created with input from psychologists and AI experts to ensure the content is safe and evidence-based. Users can type in free text about their thoughts, feelings, and questions, and the chatbot responds with supportive information based on cognitive behavioral therapy techniques. The chatbot's content is evidence-based and grounded in cognitive behavioral therapy. Researchers are exploring ways to combine different AI approaches to make responses more personalized while maintaining safety. The chatbot has been studied in multiple settings, including among college students, individuals experiencing postpartum depression, and people with substance use disorders.
  • Researchers at the University of Cincinnati and Cincinnati Children's Hospital developed a smartphone-based, AI-trained mental health chatbot to help adolescents and adults managing anxiety. The program was created with input from child psychiatrists, psychologists, and AI experts to ensure the content is evidence-based. Users interact with the tool through conversations that help them reflect on their worries, understand sources of anxiety, and learn practical coping strategies. The tool also provides guidance on how to get help. This chatbot was implemented in rural communities, specifically via MyCHN, a Federally Qualified Health Center in Texas.
  • Illinois is launching a digital maternal health pilot project that will give 56,000 people in rural Cook County access to an AI agent providing real-time maternal health advice. This project will help patients adhere to care plans, order genetic screening tests, and book vaccination appointments, among other maternal health services. Patients in the program receive a smartphone to communicate with the AI agent and a digital fitness device that tracks physical activity, heart health data, and sleep. Using each patient's personal data, the AI agent offers recommendations personalized to each patient's unique needs.

Considerations for Implementation

Conversational agents are often available 24/7, which can make it easier for patients to get information or support when needed, without the fear of stigma. However, AI conversational agents are not a substitute for trained clinicians in rural healthcare settings, and their effectiveness in addressing clinical outcomes is still emerging.

A key implementation concern is data privacy and security. Chatbot users may provide a large amount of personally identifiable information, including age, gender, and contact information — and health data, such as health conditions and medications — in an unsecure environment. There is a need for ethical frameworks to help manage privacy and data security and ensure that patients understand how their data are used.

Accuracy is another key factor. AI-powered chatbots that use large language models (LLMs) can provide personalized and conversational responses to users. However, one risk is that these systems can have "hallucinations," in which they confidently provide users with incorrect information. If the conversational agents provide incorrect responses, or responses not tailored to the patient, there may be serious implications. Specifically, there are ethical concerns around chatbots used to support people with mental health disorders, given the lack of evidence on their effectiveness.

Chatbot developers are exploring AI-augmented chatbots that combine LLMs and natural language processing (NLP) capabilities within controlled clinical settings to manage risks and build in safety features.

Another barrier to widespread implementation of digital health tools, like AI-powered chatbots, is reimbursement. Many digital health tools, particularly stand-alone apps, are not directly billable under traditional fee-for-service payment models, making it difficult for providers and healthcare organizations to sustain their use.