Regulatory Challenges of AI in Clinical Trials: Perspectives from the FDA and APAC Authorities

Regulatory Challenges of AI in Clinical Trials: Perspectives from the FDA and APAC Authorities

Introduction

The use of AI in clinical trials is tough. Using AI to analyze historical data, predict outcomes, and analyze clinical data. AI helps with patient sourcing, recruitment patterns, protocol feasibility testing, and advanced data analysis. These innovations have some challenges. For safe and ethical AI-driven trials, a risk-based strategy with open algorithms and ongoing human oversight is better than separate AI regulations. The FDA and EMA are incorporating AI into existing GCP frameworks. The FDA defines AI tools used for prediction or treatment as medical devices. But while AI is growing quickly, the rules are still catching up. This difference causes confusion for those who must meet compliance while using modern tools (Almarie et al., 2025a).

It explains the perspectives of the FDA and APAC authorities and encourages international teamwork while putting patient safety first. It also presents a path toward coordinated international efforts while giving patient safety the top priority.

The Advanced Role of AI in Clinical Trials

Clinical trials are changing because of intelligence. This change is making things quicker and easier. Artificial intelligence is also making trials more about the data. One of the changes is in finding patients for the trials. AI can do the task much more speedily as compared to old ways. Artificial intelligence can look at health records and identify people who are eligible to participate. Artificial intelligence is also being used to make the plans for the trials better. It can try out scenarios and help the people in charge make the trials more effective. Artificial intelligence is proficient at pointing out problems quickly and helps keep patients safe. Artificial intelligence predicts sequences that the human eye might overlook. In this way, it assists people to make decisions quickly. Artificial intelligence is quite difficult to use(Chopra et al.,2023).

Figure 1. Use of AI in clinical trials. AI, artificial intelligence (Chopra et al.,2023)

Hurdles on AI Regulation

The global use of AI in clinical trials has hurdles and challenges. The first hurdle is technical and data-related. The quality of AI models depends on the data used for training. If clinical data is fake and biased, the AI model generated has compromised quality. Insufficient coordination among healthcare systems makes the procedure of collecting and summarizing the large data further complicated. For this reason, rules and regulations for all organizations must be implemented so everyone can be on one page (Akinsipo & Anselm, 2025).

FDA Approach: Supporting Innovation with Oversight

Multiple devices were launched. The FDA Radiology got the most approvals, and more and more international developers got involved. Most devices utilized recently approved predicates, although predicate reuse remained uncommon. FDA initiatives, such as PCCPs and strict cybersecurity rules, improve the quality of public information. Longitudinal studies that link regulatory data to clinical and postmarked performance will be required to determine whether current pathways are adequate for the safe and equitable deployment of medical AI (Almarie et al., 2025b). Experts believe that future regulations for AI need to move away from rigid, one-time approval systems and adopt more flexible, evolving approaches. The FDA’s Center for Digital Health says that AI software needs “constant supervision.” This means regulators shouldn’t just approve it once and forget about it—they should keep checking and evaluating it as it changes over time. To support this approach, the FDA introduced a new guideline on December 4, 2024, called “Recommendations for Market Submission of Intended Change Control Plans for Software Functions of AI Devices" (Zhang et al., 2025).

The APAC Perspective: Diversity in Regulation

To create a three-level framework that focuses on identifying device features and selecting the appropriate approval process, the U.S. Food and Drug Administration (FDA) implements a system that classifies medical device software based on risk. The EU has included AIMD in structured regulatory frameworks through the MDR, raising entry barriers for market access. While making things safer, this system has led to concerns about restricting innovation. To respond, EU authorities are taking steps to simplify the regulations. Japan has developed a comprehensive policy framework for AIMD that includes rules development, research discussions, and regulatory revisions and reviews. South Korea has set up a flexible system for AIMD rules. Regulatory bodies stress the importance of improving global digital health collaboration through training programs. This will make South Korean companies more competitive on the global stage and give the country more power in international medical device regulation (Zhang et al., 2025).

Figure 2: A Comparative Review across different regions (Zhang et al., 2025)

Similar Issues Across Regions

Most areas and regions have established Active Implantable Medical Devices  (AIMD) regulations. Scholarly writing now primarily focuses on the vast problems associated with these regulations, as mentioned in the previous section. When it comes to handling the particular issues that this new technology presents, the current system still has certain shortcomings (Zhang et al., 2025).

The Path Toward Harmonization

To fully leverage intelligence in clinical trials, it is essential to establish uniform rules across all locations. Globally, organizations are trying to make such rules that everyone can follow. We require collaboration between rule-makers, companies, and universities to ensure uniformity. This coordination is important for setting standards that work for everyone. Some countries are also testing out different ideas to see which ones work best. (2015–2025) Patient safety is the first priority in mind as we work on these issues. Ensuring the use of artificial intelligence is crucial. These rules are really the best. Artificial intelligence will help people to trust intelligence over time from 2015 to 2025. Artificial intelligence can act strangely in tests. We need to be careful when we use AI. It is important to use artificial intelligence in a moral and safe way. In the future, rules will pay more attention to ethics. This will ensure that artificial intelligence in devices is not just safe and effective but also meets patients' needs. We need to use intelligence in a way that respects rights and patient well-being. The thing is it is not about how well the technology works for the patient. We have to think about what's best for the patient when we use intelligence. Patient rights and patient values and patient well-being are very important to us.

Experts in the field say that artificial intelligence can cause problems if it is not properly regulated and watched over. Artificial intelligence can provide us results, put patients' private information in danger, and use their sensitive health information in a detrimental way. After a while, these problems can make people trust artificial intelligence less in healthcare, which means that patients and doctors will not want to use it much. Overall having rules about what is right and wrong will help us make a healthcare system where artificial intelligence helps doctors make new discoveries while also keeping patients safe and making sure people trust it. Many countries are making new rules to address problems related to artificial intelligence. The primary objective is to ensure the development and use of artificial intelligence from the outset. For instance, the European Union's Artificial Intelligence Act says that artificial intelligence systems need to be trustworthy, easy to understand, and fair to everyone. Having rules in place will help make healthcare better and also keep patients safe and make them trust.

Conclusion

Clinical studies take a lot of physical effort and are long and complex. Consequently, healthcare companies have been looking to artificial intelligence (AI) to assist in solving issues connected to gathering vast volumes of data, patient recruiting, remote tracking, and patient engagement about their participation in the trial. It can create unique solutions from unorganized or electronic data; artificial intelligence will enable faster and more organized execution of clinical trials. Clinical studies take longer under conventional procedures since recruiting patients, enrolling them, often evaluating patient development, and verifying patient compliance with prescribed treatments all contribute to their length. Still, artificial intelligence technologies let one automate significant portions of the tasks listed above. Along the duration of the trial, artificial intelligence will not only support healthcare professionals in gathering patient data but also organize it. Furthermore, artificial intelligence will enable smooth contact with other information systems and aid in producing automated trial reports by means of efficient analysis of gathered data. On the whole, artificial intelligence enables a more effective approach to running a clinical trial. Furthermore, other advantages from better data collection, bio-simulation, and early disease detection are currently accessible artificial intelligence tools. Hence, present data points to the possibility that AI can greatly lower the cost and duration of a clinical study and enhance the effectiveness and accuracy of the procedures for developing new medicines. AI will keep growing its impact on the conduct of clinical trials in the following years and help to improve medical research outcomes even more.

References

Akinsipo, O. B., & Anselm, O. H. (2025). Challenges and opportunities for implementing green chemistry in Nigerian universities: Educational and policy perspectives. Sustainability & Circularity NOW, 02(CP), a-2534-1903. https://doi.org/10.1055/a-2534-1903

Almarie, B., Gonzalez-Gonzalez, L. F., Dos Santos Barbosa, L. A., Lutz, A., Grosse, U., & Fregni, F. (2025a). Machine learning-enabled medical devices authorized by the US Food and Drug Administration in 2024: Regulatory characteristics, predicate lineage, and transparency reporting. Biomedicines, 13(12), 3005. https://doi.org/10.3390/biomedicines13123005

Almarie, B., Gonzalez-Gonzalez, L. F., Dos Santos Barbosa, L. A., Lutz, A., Grosse, U., & Fregni, F. (2025b). Machine learning-enabled medical devices authorized by the US Food and Drug Administration in 2024: Regulatory characteristics, predicate lineage, and transparency reporting. Biomedicines, 13(12), 3005. https://doi.org/10.3390/biomedicines13123005

Chopra, H., Shin, D. K., Munjal, K., Dhama, K., & Emran, T. B. (2023). Revolutionizing clinical trials: The role of AI in accelerating medical breakthroughs. International Journal of Surgery, 109(12), 4211–4220. https://doi.org/10.1097/JS9.0000000000000705

Zhang, S., Li, Y., Liu, W., Chu, Q., Wang, S., Li, J., & Chen, Y. (2025). A decade of review in global regulation and research of artificial intelligence medical devices (2015–2025). Frontiers in Medicine, 12, 1630408. https://doi.org/10.3389/fmed.2025.1630408