A new framework for the safe use of artificial intelligence in healthcare

By publishing a new framework for monitoring agent-based artificial intelligence in the health system, the American Medical Association emphasized the necessity of evaluating functional evidence, determining the limits of authority, effective monitoring and determining the responsibility for possible errors of this technology in patient care.

According to Ambassador Arjan, quoted by healthcaredive, the American Medical Association (AMA) has published a new framework for monitoring the safe and effective use of artificial intelligence agents in the health system. This framework emphasizes the principle that agent-based artificial intelligence solutions for performing clinical tasks must be supported by valid evidence and provide high-quality and reliable performance.

This document raises four basic questions that are related to determining the authorized tasks of artificial intelligence, the conditions of its use, and the responsibility of individuals and organizations for the performance of these tools.

Artificial intelligence agents can automatically perform complex and multi-step tasks; Capabilities that go beyond the usual ability of productive artificial intelligence tools to produce text and content. These technologies are increasingly being used in clinical care, including automated documentation of medical sessions, assisting in the diagnosis of diseases and monitoring the condition of patients.

However, the operational independence of these tools also brings risks. Since AI agents can perform several steps of a process without direct human supervision, their incorrect or biased decision may cause significant harm to the patient before the physician intervenes.

In its new framework, the American Medical Association has asked four questions to evaluate artificial intelligence agents used in clinical activities:

1. What clinical task does artificial intelligence perform? It should be clear what the tool is designed to do and what exactly it is responsible for.

2. What evidence supports AI's ability to perform this task, and what happens if it fails? Assessing performance evidence and predicting the consequences of errors is an important part of the monitoring process.

3. What actions is artificial intelligence allowed to perform and under what conditions can it perform them? Determining the limits of the tool's authority and the conditions of its use is necessary to reduce possible risks.

4. Who is responsible for the risks and consequences of AI performance? Accountability for the decisions and results of the performance of these tools should be clearly defined.

"As AI capabilities advance, evidence, authority, oversight, and accountability must continue to align to serve to improve patient care," the framework states.

The American Medical Association plans to use this framework in collaboration with other organizations to ensure the safe and responsible use of artificial intelligence agents in patient care activities. This approach also emphasizes the central role of doctors in the development and application of these technologies.

The publication of this framework takes place in a situation where the use of agent-based artificial intelligence tools in the health sector is increasing rapidly.

According to a recent survey of 250 healthcare executives in the United States, about 28 percent of respondents said their organization had started using these tools. Also, 44 percent said they are running pilot projects or proof-of-concept projects.

Most respondents expect agent-driven artificial intelligence to have a transformative impact on clinical work processes.

However, 72 percent of respondents reported that AI tools are sometimes implemented in their organizations without IT approval; An issue that indicates the existence of a gap in monitoring the use of this technology.

Lack of adequate monitoring can create important challenges for health system managers. More than half of the respondents cited the possibility of violating legal and regulatory requirements and unauthorized access to protected health information as their main concerns.

Developments in the field of artificial intelligence in health are progressing at such a speed that laws and regulations can hardly keep up with it. In a situation where comprehensive and effective regulations for this technology have not yet been formed at the federal level of the United States, the specialized groups and trade unions of the health industry are trying to cover part of the regulatory gap by compiling guidelines and setting executive limits.

At the state level, a scattered set of laws and regulations is being formed. Some states have imposed restrictions on the use of AI in health care, while others are pursuing pilot AI-based care initiatives.

According to the information published by Manet legal and consulting company, in 2025, more than 250 bills related to artificial intelligence and health were presented in 47 American states.

These developments show that with the expansion of the use of artificial intelligence in treatment decisions and processes, determining the limits of the authority of this technology, monitoring its performance and determining the responsibility for possible errors has become one of the important issues of health policy.

A new study of 8,783 Canadian adults found that those who ate the most highly processed foods were about 40 percent more likely to have signs of silent brain damage; Changes that may occur without detectable problems in memory and thinking.

Both magnesium and vitamin B12 play a role in the normal functioning of the body and processes related to energy production; But while magnesium helps nerve and muscle function, vitamin B12 is essential for the production of red blood cells, and its deficiency can lead to anemia and fatigue.

In the cold seasons, when persistent coughs after respiratory infections become a common problem, the medical community has taken a special look at ancient and complementary solutions to reduce throat irritation.

Fatty liver has become one of the most frequent medical diagnoses in Iran. From periodic ultrasounds at work to simple visits to the general practitioner, the term "grade one fatty liver" or "grade two" has almost become a part of people's daily conversation. In the same space, home and traditional recommendations have become bolder, and one of the most frequent ones is the consumption of dill, especially in the form of tea, powder, or adding it to yogurt and salad.

Choosing the right foods for breakfast can be a simple way to support cardiovascular health. Nutritionists recommend including foods such as beans, avocados, potatoes, bananas, and kiwis in your diet to increase potassium intake.

Women's health experts say it's time to put aside 6 common beliefs about menopause and get to know your body better, and go through this natural stage of life more consciously.

If you experience heaviness or digestive discomfort after consuming some protein foods, it is important to choose the type of protein and its preparation method. Eggs, chicken, tofu, and Greek yogurt are among the options that are usually easier to digest and can be part of a balanced diet to support gut health.