AI Agents in Healthcare.

Core concepts and methodologies that form the basis of AI systems include:

·        Natural Language Processing (NLP) - Extracting clinical concepts such as

symptoms, diagnoses and treatments from narrative text, such as clinical notes

·       Classifiers - Mapping input data into categories or classes based on training data

in which the proper classification is known, i.e. labelled, so new data can be correctly categorized

·       Artificial Neural Networks - Systems modeled on biological nervous systems

·       Machine Learning (ML) - Systems able to process large volumes of data and extract meaningful information from it to address practical problems

·       Augmented Intelligence - Technology intended to assist humans in utilizing or extending their own capabilities

·       Image/Speech Analysis - Extracting meaningful information from images such

·       as MRIs or recorded speech, as opposed to numeric, categorical or text data

At the beginning of August 2024, the Artificial Intelligence (AI) Act came into force, which imposes a number of obligations on those marketing or commissioning an AI system under their own name.

The application of artificial intelligence (AI) in medicine raises at least as many hopes as concerns. On the one hand, there is the promise of faster, more accessible and cheaper diagnostics; on the other hand, there is concern about misdiagnosis of patients' conditions. The prospect of surgery performed by an AI-equipped robot, in view of the low rate of doctors per 1,000 inhabitants in Poland [according to an OECD report, in 2021 there will be 3.4 doctors per 1,000 inhabitants in Poland], fills us with optimism, nullified by the risk of a medical error and the lack of clear regulations on liability for harm or other types of damage. Examples could be multiplied.

At the same time, the spread of the use of AI in medicine seems to be inevitable. It is therefore worth examining the most relevant regulations in this context, especially as the Artificial Intelligence Act (AI Act) entered into force at the beginning of August 2024.

The AI Act imposes a number of obligations on suppliers, i.e. entities marketing or commissioning an AI system under their own name, the most important of which relate to issues such as:

- the obligation to implement a high-risk AI system risk management system;

- recording of events of high-risk AI systems;

- the provision of human oversight of the high-risk AI system;

- the obligation to ensure transparency of the AI system that interacts with humans;

The AI Act also distinguishes - in addition to the supplier - the other actors involved in the operation of AI systems in the economy, i.e. users, applicators, importers, distributors and operators, who are also subject to obligations relating to their role in the introduction of an AI system into the economy or in the use of AI systems.

These obligations aim to create a comprehensive framework that ensures accountability and safety throughout the lifecycle of AI systems, fostering trust among users and stakeholders alike.

It is first necessary to identify which entity related to the operation of AI systems in business, as defined in the AI Act, is the entrepreneur. Next, it is necessary to determine whether the AI system used by the trader is a high-risk AI system.

The above qualification will determine the scope of the obligations under the AI Act.

Failure to comply with the prohibition of AI system practices or non-compliance of the AI system with the requirements, explicitly indicated in the AI Act, may result in an administrative monetary penalty for the trader. In addition, traders must ensure that they have implemented adequate risk management measures and maintain thorough documentation to demonstrate compliance, as regulatory bodies will likely scrutinize these aspects during audits. Traders should also stay informed about any updates to the AI Act, as regulatory frameworks are continually evolving to address emerging technologies and their implications.