AI Agents for Healthcare: Benefits and Use Cases

AI agents, which are digital assistants designed to help organizations automate certain kinds of workloads and improve decision-making, have the potential to help change the way healthcare is delivered and how people manage their health.

Although healthcare organizations are beginning to adopt AI for isolated use cases, the real value will be realized with the adoption of AI agents that are each responsible for different tasks but work together seamlessly. In a clinical setting, one AI agent might be trained to record and interpret verbal instructions or conversations between a physician and patient, another to understand lab results, and a third to code a treatment plan for proper reimbursement. Working together, AI agents can help create a big-picture view of patients that physicians can use to make better-informed care decisions.

What Are AI Agents?

Broadly, AI agents, combined with a variety of data types, use large language models (LLMs—machine learning models that can perform natural language processing tasks) and retrieval-augmented generation (RAG—which provides a way to optimize the output of an LLM without modifying the model itself). AI agents can be assigned tasks, examine their environments, take actions based on their roles, and fine-tune their behavior based on their experiences and user feedback. These tasks can range from answering simple questions and analyzing language context and tone to resolving complex operational challenges in industries such as healthcare, retail, and hospitality.

What Are AI Agents in Healthcare?

In healthcare settings, using chat, text, or voice interfaces, AI agents can summarize spoken words, uncover signals that require human attention, and scan internal and external data to provide users—patients and clinicians—with real-time results and continually improve performance and accuracy. AI agents work by taking in human language requests, encoding them, and then sending them to the enterprise data store. The agent uses an LLM to understand the query, then searches the knowledgebase for relevant data, re-ranks the content for semantic relevance, combines the most relevant content and the query into a coherent response, and then sends the response and the content used to create it to the requester.

At a basic level, by automating routine tasks, AI agents in healthcare use artificial intelligence to help reduce the workloads of healthcare professionals, including administrators, allowing them to focus more on patient interactions, higher-level decision-making, and operational improvements.

At a more advanced, clinical level, AI agents can analyze vast amounts of data from EHRs, medical research repositories, government regulation libraries, and other sources to assist in diagnoses and help personalize treatment plans based on patient histories and other characteristics. They use predictive analytics to process and interpret large data sets, both historical and current, to help providers make more-informed choices and improve patient outcomes.

For example, an agent might take historical data about disease outbreaks and interpret it against current laboratory value patterns. If it identifies a cluster of certain values in a community, it can provide clinicians with insights and predictions pertaining to a potential outbreak. The agent tasked with parsing current data sets may call upon another agent for regional or national data and would need to know the values present at the baseline and what the break point would be for an outbreak. Another set of agents might be directed to pull together data from a repository of mammograms, which would show the typical progression of disease associated with a particular type of mammogram, compare that data set with a single patient’s mammogram, and in doing so help determine how a condition may progress and what options the physician has for treatment.

Key Takeaways

• Physicians continue to report high levels of stress related to the volume of administrative work they’re expected to accomplish. AI agents can help alleviate much of this stress by acting as digital assistants that are integrated into the organization’s EHR system.
• Healthcare AI agents can support a more cost-efficient, productive operation by streamlining and automating administrative processes.
• Although an AI agent may appear to the user as a single entity, behind the scenes there may be dozens of agents tasked with responsibilities as varied as analyzing lab test results, managing prescription refills, and providing a physician with specific recommendations related to a patient’s condition.

AI Agents for Healthcare Explained

AI agents for healthcare are digital assistants that help improve interactions between patients and healthcare providers by combining healthcare intelligence with voice, chat, and text messaging interfaces.

Rather than acting as standalone applications, healthcare AI agents are embedded into administrative and clinical workflows, such as patient registration, during which an agent can automate the filling out of lengthy, repetitive forms. Physicians can call on agents before appointments and request a “pre-briefing” on a patient through their device to acquaint themselves with the patient’s medical history, most recent test results, and reason for their visit while walking to the exam room. (That’s an example of a lead agent pulling together information from other, specialized agents to create a unified report.) Physicians can also call on AI agents during appointments, when, with the patient’s permission, an AI agent can listen in on patient-physician interactions and help automatically create a summary of what was discussed and decided on. Healthcare AI agents can also be trained on data sets that are specific to a condition or disease so physicians can benefit from the broadest range of clinical knowledge when treating their patients.

In practice, the biggest value of healthcare AI agents, at least initially, will likely come from their ability to take manual data entry out of the hands of clinicians so they can focus on the patient, applying their medical knowledge and human intuition. Although physician burnout is on the decline since the pandemic, according to the American Medical Association, almost half of physicians still report at least one symptom. One stressor is the volume of administrative work they’re required to perform.

Another benefit of AI agents is that they can help align treatment codes with payer guidelines so practitioners are reimbursed for the care they provide. It’s an important consideration given that US healthcare organizations operate with an average profit margin of just 4.5%, according to the Kaufman Hall National Hospital Flash Report published in November 2024.

AI agents require intensive computing power, far more than any healthcare organization would have onsite, so running them in the cloud is a necessity. The cloud also gives healthcare organizations the benefits of large language models trained on medical data sets. Alternatively, that training can take place on private data sets in a private cloud so organizations don’t lose control of their own data.

The Importance of AI Agents in Healthcare

The healthcare industry is still in the early stages of adopting AI agents due to the industry’s complexity and regulations that govern how they can be used. For example, a task such as renewing a prescription might seem ripe for automation, but the agent would first need to factor in whether it’s safe to provide additional doses of the medication without a recent in-clinic examination or telemedicine appointment. However, when implemented correctly, AI agents have the potential to free clinicians from data entry tasks, help improve care reimbursement rates and accuracy, and help improve healthcare decision-making. (More on these and other benefits later.)

How Do AI Agents in Healthcare Work?

Healthcare organizations can use a combination of agents, rather than one single agent. Each agent is designed to perform a specific task, such as setting up appointments, preregistering patients, prepping clinicians, recording and summarizing details of an examination, and managing patient follow-up. They work by tapping into a vast store of knowledge from internal sources (including patient EHRs) and external sources to recognize patterns and understand user needs.

Here’s the standard planning process for creating a set of AI agents:

1. The agent is programmed according to organizational objectives, which are prioritized so critical actions take precedence.
2. The agent is trained on the internal and external data needed to manage requests, converting raw data into usable data.
3. The agent executes in the service of organizational goals by allocating tasks to different agents. Agents monitor their own accuracy and track actions and their outcomes. They modify tasks as needed based on human feedback and changing circumstances.
4. The agent measures success against the original goals, identifies areas for improvement, and incorporates new information and human feedback into the knowledgebase to continuously improve its performance.

How Are AI Agents Transforming Healthcare?

The short answer: They’re not yet. As stated earlier, the industry is still in the early stages of adopting AI agents. And although patients will perceive AI agents as a single source of help and knowledge, in actuality there are many different agents, all tasked with handling different parts of the healthcare journey, and these agents will likely come from multiple vendors, all with expertise in specific areas.

That said, AI agents do have the potential to help transform the healthcare industry, as they can relieve clinicians of many of their manual data entry tasks and help them gain a more informed and focused view of patients while providing patients with a trusted “assistant” to help them navigate the complex healthcare system and achieve better health outcomes.

Benefits of AI Agents in Healthcare

AI agents stand to benefit healthcare organizations and their patients in two main ways: by helping to improve clinical decision-making and treatment and by reducing the cost and burden of administrative tasks. For more on these and other benefits, read on.

• Support for healthcare providers. AI agents can assist physicians with decision-making by providing them with patient history summaries prior to appointments and access to machine learning tools trained on specific clinical data sets. For example, an oncologist with a lung cancer patient might ask the AI agent to pull together data from various locations—for example, the latest clinical research on the condition, lab reports, CT scans, and the patient’s self-reported lifestyle habits—and provide predictive modeling analysis that can aid them in recommending a course of treatment. Agents can also help reduce clinician and administrator burnout by eliminating certain manual data entry tasks.
  
  Community hospital St. John’s Health is using AI agents to make it easier for physicians to stay up to date with post-visit notes on patients. Physicians are able to walk into an exam room with their mobile device and, with the patient’s permission, set it to ambient listening. The agent sifts through the conversation and identifies the important information for continuity of care and billing, and then puts it into a concise, clean digital summary.
• Cost reduction. With operating margins that can run lower than 5%, healthcare organizations need to tightly control costs. By using AI agents to automate and streamline billing, coding, and payer reimbursement processes, organizations can lower their administrative expenses without sacrificing care quality.
• Improved diagnostics. AI agents can help provide clinicians with concise summaries of patient medical history (and even genomic) data, relevant medical research, and data gathered from patient medical devices, as well as analyses of X-rays, CT scans, and MRIs, to support more-accurate diagnoses.
• Personalized treatment. Using patient data from multiple sources, AI agents working in concert can help generate personalized treatment plans for review and approval by clinicians. AI agents can also take in sensor data from personal medical devices and alert care providers when readings are out of range.
• Enhanced efficiency. On average, physicians spend about 15 minutes with a patient, and they need another 15 to 20 minutes to update the patient EHR. By automating EHR updates and the coding of treatments to facilitate accurate reimbursement, AI agents can help doctors gain back time to spend with patients and work with extended care teams to make clinical decisions.
• Real-time monitoring. By connecting to remote patient monitoring tools, such as smartwatches, heart monitors, and glucometers, AI agents can continuously track patient health. As a result, rather than just relying on information gathered during office visits or sudden ER trips, doctors have access to a steady stream of health data that can be parsed and interpreted by the AI agent. Doctors can receive only the alerts that require their intervention. This kind of real-time monitoring can also help patients become more invested in their own health status, as the AI agent can be set up to communicate with them using natural language.
• Faster drug development. It’s impossible for doctors to know about every clinical trial that might be useful for their patients. AI agents will be able to continually track clinical trials and alert doctors when one is available that fits a specific patient’s set of conditions and medical history. This has the potential to accelerate research and the discovery of new treatments.
• Increased accessibility. Conversational AI agents use natural language to make it easier for patients to take control of their healthcare—for example, patients can query an agent about a symptom, alert an agent to schedule an appointment, or get reminders from an agent about prescription renewals.
• Predictive insights. AI agents that use predictive data analytics could potentially make it easier for doctors to predict patient medical conditions and health risks so they can tailor treatment plans.

Components of AI Agents in Healthcare

AI agents use a variety of inputs to do their work, depending on the agent’s specific role in the healthcare environment. A patient-interaction agent will converse with patients, tapping different sources of patient and other data to respond to queries and provide lifestyle support. Another agent will listen in on an examination and pull out the specific information needed to update the patient record. Some agents respond only to requests from other agents—for example, an agent responsible for summarizing a treatment plan would connect with an agent that understands laboratory values and another that can interpret radiology images. Most healthcare AI agents require a complex mix of the following components:

• Perception. Healthcare AI agents use audio and video recorders to gather information from their environment—for example, a physician’s office—and convert that data into formats that can be input into a patient’s EHR record and used by a physician to recommend a course of treatment.
• Action. This component takes action based on agent analyses, insights, and directives. Agents can interact with their environments—for example, by providing physicians with a visit summary, suggesting diagnosis options, and recommending treatments that take into account the entirety of a patient’s history. Additionally, an agent can chat with a patient outside the treatment location to provide medication reminders or healthy lifestyle suggestions.
• Utility. To determine the efficacy of an AI agent, providers can measure how well the agent achieves its goals. Criteria can include patient outcomes, user satisfaction, and the degree of accuracy of its clinical recommendations.
• Learning. AI agents can use human feedback loops to help improve their results—for example, by noting which sets of proprietary data and questions produce the best results. Agents can also become better at tasks over time by receiving additional training on new data sets. When physicians validate the analysis or recommendations that the AI agent makes, the agent can use that data to guide future actions.
• Reasoning. Using both acquired and stored data, AI agents can apply reasoning techniques to help interpret the data, predict the likelihood of certain outcomes, and provide options that clinicians can use to make informed decisions about the best course of action for a patient.
• Memory. An AI agent’s memory module not only stores patient data and medical research, but it also learns from user feedback to continuously improve care recommendations over time.

Key Use Cases of AI Agents in Healthcare

The best healthcare use cases for AI agents take advantage of AI’s ability to analyze vast amounts of data to help improve patient care and reduce administrative overhead. The best use cases are ones that enable the agent to learn over time. The following rise to the top:

1. Diagnostic support. AI agents can aid clinicians in providing more accurate diagnoses by analyzing medical data—including lab results, digital scans, patient histories, and medical literature—to assist in identifying conditions and diseases.
2. Treatment recommendations. AI agents can help physicians tailor treatment plans based on patient needs and the latest medical research, clinical guidelines, and best practices. These are just recommendations. Considering the agent recommendations, physicians need to choose the path they think will produce the best outcomes based on their experience and judgment.
3. Predictive analytics. By factoring in data such as patient age, gender, geography, lifestyle, health history, and even genomics, predictive analytics can help forecast disease risks and patient outcomes.
4. Medical imaging analysis. Agents can help detect anomalies in X-rays, MRIs, and CT scans and automatically send images to specialists for review, facilitating improved diagnostic accuracy.
5. Clinical decision support. It would take 13 years for a physician to read all the medical literature coming out in a single year, according to one estimate. That’s why clinicians stand to benefit from generative AI agents that can ingest a vast amount of research data, summarize it, and deliver relevant recommendations based on a patient’s health status.
6. Drug discovery. Pharmaceutical companies are using AI to help sift through chemical compound libraries and summarize health science literature, clinical trial data, patient profiles, and even patient demographic information. AI agents can step in to accelerate the identification and development of the most promising new medications, even ones that have previously been discounted by human researchers as treatments for certain conditions.
7. Patient monitoring. AI agents can track health data from patient wearable devices and other at-home medical equipment to produce real-time alerts when blood pressure, glucose, and other variables reach certain levels. Even more important, they can sift through the huge volumes of data that many medical devices generate to give healthcare providers only the information they need to improve patients’ short- and long-term care.
8. Virtual health assistants. AI-powered virtual health assistants can help answer patients’ queries, provide them with health guidance and reminders, and track their health data. Generally available through a mobile app or website, the tools can communicate with patients through natural language interfaces and access large health data sets to provide patients with answers that are accurate and up to date.
9. Administrative automation. AI agents can help streamline appointment scheduling, billing, and recordkeeping, as well as provide immediate responses to patients’ administrative inquiries and manage their prescription refill requests. They can also automate patient onboarding, freeing up staff to focus on more crucial issues.
10. Mental health support. AI-supported therapy apps are being used to help treat depression and anxiety. The AI agents “converse” with app users, asking questions as a therapist would and identifying language in the responses—words and phrases—that can signify mental health problems and help patients recognize their emotions and adopt techniques that reduce negative feelings. Mental health agents are goal-driven and can autonomously use human feedback to make decisions about the best way to meet those goals—for example, to help a patient stabilize or reduce thoughts of self-harm or dependence on alcohol. This type of support can be particularly valuable in regions that lack adequate mental healthcare options.

Future of AI Agents in Healthcare

Most of the early technologies used in healthcare were cumbersome, time-consuming, and frustrating for clinicians at every level. Clinicians were responsible for knowing where different sets of data were stored and for pulling together that data to obtain a complete, accurate view of the patient. Relieving that administrative burden has been the highest priority for healthcare technology companies, since that burden has contributed to burnout, early retirement, and physicians simply leaving the profession for other, less stressful positions. AI agents can promise to help relieve that burden, as well as reduce diagnostic errors, letting physicians engage more with patients, improving health outcomes, and helping ensure physicians are paid in a timely manner for the services they provide.

Elevate Patient and Clinical Experiences with Oracle

With the acquisition of EHR developer Cerner, Oracle has expanded its already substantial portfolio of healthcare technology products and services, which is augmented by its deep expertise in data management, cloud applications and infrastructure, and AI. Oracle Health Clinical AI Agent can help assist at every point in the patient lifecycle, from initial intake to clinical follow-up. By automating the entire documentation process and synchronizing data with the patient’s EHR, the AI agent can contribute to a better patient experience and help improve diagnoses and treatments.

AI Agents for Healthcare FAQs

How will AI be used in healthcare?
Although there’s no substitute for the knowledge, experience, and intuition of a talented healthcare practitioner, AI can become a trusted assistant by automating scheduling, check-in, and other administrative processes, summarizing the details of examinations, helping inform diagnoses and treatments, and helping manage follow-ups with patients.

Which AI tool is used in healthcare?
The most common AI tools so far are GenAI-based assistants that can help improve the accuracy and speed of documentation and reduce clinician administrative burdens.

What are the types of AI agents?
Types of AI agents include simple reflex agents, which react to input without considering the broader context; model-based reflex agents, which use a model of the environment related to their function to assess the effects of actions before making a recommendation; goal-based agents, which consider long-term objectives and make recommendations based on that information; utility-based agents, which perform a single function; and learning agents, which adjust their performance over time based on interactions with users.

What is the most common AI in healthcare?
Although AI is beginning to take on many functions in healthcare organizations, some of the most common are analyzing lab results, summarizing appointments, and interpreting paper forms and scanned images such as X-rays and CT scans.