ARTIFICIAL INTELLIGENCE IN CLINICAL TRIALS – PRACTICAL AND LEGAL ASPECTS
Publication date: April 16, 2024
Artificial intelligence (AI) is a field of science that deals with the creation and use of computer systems that can simulate human cognitive processes, such as learning, reasoning, decision-making and problem solving. AI has an increasing impact on various aspects of our lives, including medicine and health. This article will discuss how AI changes clinical trials, i.e. the process of testing new drugs and therapies on humans before they are approved for marketing and use.
Application of AI in clinical trials
Clinical trials are necessary to test the effectiveness and safety of potential drugs that may help treat or prevent various diseases. However, clinical trials are also difficult, long-lasting, very expensive and carry a high risk of failure. According to some estimates, only a few percent of drugs that begin clinical trials make it to the registration and approval phase[1]. Therefore, many pharmaceutical and biotechnology companies are looking for ways to improve the efficiency and quality of clinical trials, and one of them is the use of artificial intelligence. Among the many examples of AI applications in clinical trials, the most important is the possibility of designing and planning trials using artificial intelligence. AI can help select appropriate therapeutic targets, design new drug molecules, predict pharmacological and toxicological properties, and optimize research protocols and criteria[2]. Thanks to the ability to operate on huge databases and analyse all information at once, AI can accurately indicate the direction of research undertaken by pharmaceutical companies, what is more, artificial intelligence, by interpreting the entered data, can detect certain patterns, correlations and anomalies in the data that a human would not be able to notice due to the extensiveness of the analysed data[3]. Another application for artificial intelligence in clinical trials is patient recruitment and monitoring. It should be noted that this area of clinical research – recruitment – enjoys exceptional interest, because in a study conducted by the Massachusetts College of Pharmacy and Health Sciences over the last 5 years, among the articles analysed, over 50% concerned recruitment as a possibility of using AI (artificial intelligence) in clinical trials[4]. Artificial intelligence based on patients’ demographic, medical and genetic data could significantly help identify and select patients for appropriate trials, which would increase the likelihood of research success. Moreover, thanks to the use of modern wearable devices (e.g. smartwatches) and mobile applications, artificial intelligence could monitor the health status and response of patients to treatment in real time, which would translate into comprehensive collection of data needed to conduct clinical trials. Such a procedure would improve the quality of data and simplify its collection – patients would not have to devote their time to providing the necessary data, they would only have to wear the indicated technology collecting data on their biological parameters[5]. Currently, there are many companies that actively use artificial intelligence for their purposes, one of them is BenevolentAI, which uses AI to analyze huge amounts of scientific, clinical and biological data to accelerate research on new drugs, including: in the treatment of dengue[6]. Another example of a company using AI in clinical trials is Science 37, in the case of this company, artificial intelligence helps recruit patients from various sources who are very well matched to specific studies, thus speeding up the entire recruitment process while increasing diversity among patients[7].
Limitations in the use of AI in clinical trials
Due to the use of artificial intelligence in clinical trials, in addition to many advantages and benefits, there are also certain limitations, problems and controversies associated with the use of AI. First of all, technical and methodological issues should be mentioned – the tool of artificial intelligence undoubtedly requires a huge amount of data on which it will “learn” to function properly. If the database is not complete or is of low quality, the effects of the AI’s work may be burdened with imperfect answers, i.e. it may be biased towards certain phenomena and issues, respond imprecisely in terms of the information generated, because “the algorithm is only as good as the database data on which it was trained”[8], which in turn raises the problem of how to create, if at all possible, a complete, adequately representative and high-quality database. Another controversial issue emerging in the public space is the ethical and legal issue, there is a fear of abuse in connection with the use of AI in the aspect of security and privacy of patient data. AI tools without appropriate legal regulations may create a lack of trust and responsibility if patient data is not properly protected. The described problem is a challenge both for entities providing services using artificial intelligence, which must create appropriate mechanisms to guarantee the transparency and security of AI operation, and for relevant authorities, which must take appropriate legislative steps and create a legal framework regulating the use of new technologies in the medical aspect, in including in clinical trials. Another problem, strongly related to the previous issue regarding the lack of appropriate regulations, is the issue of liability for errors. At the moment, there is no clearly defined method of determining liability as a result of an error committed by the AI model, there is a strong need to regulate this issue[9].
Regulatory status of artificial intelligence in clinical trials
The problem of the lack of appropriate legal regulations regarding the use of AI in clinical trials is not ignored, it has been noticed both internationally and in individual countries. At the national level in Poland, there are no dedicated regulations regarding AI in medicine, including clinical trials, legislative work is undertaken primarily at the EU level, because the acts adopted so far, although they cover the issues of new technologies, do not directly refer to the use of artificial intelligence in clinical trials[10], these include: data protection regulation (GDPR), the AI Act, or the MDR regulation (Medical Devices Regulation), although the provisions contained in these acts can be applied as auxiliary measures, within the European Union it is also planned to create a specialized act regarding the use artificial intelligence to support the development and use of medicines. Such a project is currently being dealt with by the European Medicines Agency, which is conducting extensive public consultations in connection with the publication of “Reflection paper on the use of artificial intelligence in the lifecycle of medicines”[11]. The document discusses the principles relevant to the use of AI at every stage of a drug’s life, from discovery to approval. The topic of AI regulation in the United States is also widely discussed, but they present a less formalized approach than the EU, they are based primarily on the guidelines issued by the FDA (Food and Drug Administration), which sets the rules and framework in which to operate when it comes to the use of AI in clinical trials. Above all, the FDA emphasizes data quality, human-led project management, and standards in the development of AI models[12].
Summary
Models based on artificial intelligence are an amazing tool that can significantly support human work in clinical trials, however, they require attention and the creation of appropriate legal and ethical frameworks by relevant authorities in cooperation with scientific institutions, social organizations and entities providing AI services. Artificial intelligence is undoubtedly an interesting direction of development for the pharmaceutical industry and even for medicine as a whole, as AI-based tools offer a very wide range of application possibilities, although in order for humanity to fully benefit from the potential of artificial intelligence, solid legal, ethical and security foundations are needed as well as transparency of action.
[1] Szłapka P., Artificial intelligence in clinical trials, https://pacjentwbadaniach.abm.gov.pl/pwb/aktualnosci/aktualne-wydarzenia-ii/1791,Sztuczna-inteligencja-w-badaniach-klinicznych.html (access: 11/01/2024).
[2] Askin S., Burkhalter D., Calado G., El Dakrouni S., Artificial Intelligence Applied to clinical trials: opportunities and challenges, https://link.springer.com/article/10.1007/s12553-023-00738-2 (access: 11/01/2024).
[3] Anagnostopoulos C., Champagne D., Devenyns T., Deverson A., Tarkkila H., How artificial intelligence can power clinical development, https://www.mckinsey.com/industries/life-sciences/our-insights/how- artificial-intelligence-can-power-clinical-development (access: 11/01/2024).
[4] Askin S., Burkhalter D., Calado G., El Dakrouni S., Artificial Intelligence Applied to clinical trials: opportunities and challenges, https://link.springer.com/article/10.1007/s12553-023-00738-2 (access: 11.01.2024).
[5] Mai B., Roman A., Suarez A., Forward Thinking for the Integration of AI into Clinical Trials, https://acrpnet.org/2023/06/forward-thinking-for-the-integration-of-ai-into-clinical-trials/ (dostęp:11.01.2024).
[6] https://www.benevolent.com/news-and-media/press-releases-and-in-media/dndi-and-benevolentai-collaborate-accelerate-life-saving-drug-discovery-research-dengue/ (access: 11.01.2024).
[7] https://www.science37.com/patient-recruitment (access: 11.01.2024).
[8]Melke M., Greser J., AI in clinical trials: a chance for development? , https://www.polon.pl/technologia/ai-w-badaniach-klinicznych-szansa-na-rozwoj/ (access: 12/01/2024).
[9]Srivastava S., The Future of Clinical Trials – Unlocking AI’s Potential to Revolutionize Healthcare Research , https://appinventiv.com/blog/artificial-intelligence-in-clinical-trials/ (access: 12/01/2024).
[10]Melke M., Greser J., AI in clinical trials: a chance for development?, https://www.polon.pl/technologia/ai-w-badaniach-klinicznych-szansa-na-rozwoj/ (access: 12/01/2024 ).
[11] https://www.ema.europa.eu/en/news/reflection-paper-use-artificial-intelligence-lifecycle-medicines
[12] Licholai G., AI In Clinical Research: Now and Beyond, https://www.forbes.com/sites/greglicholai/2023/09/18/ai-in-clinical-research-now-and-beyond/?sh=500ded473c85 (access: 12.01.2024).
