Can AI or ML Help You Design a Better Clinical Trial?

The integration of Artificial Intelligence (AI) and Machine Learning (ML) in clinical trial design is beginning to reshape the pharmaceutical and medical research industries. Our article explores the methods by which AI and ML are modifying clinical trial design, with a focus on recent advancements and methodologies endorsed by regulatory bodies and used by industry leaders.

AI and ML in Clinical Trial Design

Traditional clinical trial processes can be slow, costly, and burdened by inefficiencies. The use of AI and ML is seen as a solution to many of these challenges, particularly in the design phase of trials where decisions about patient selection, trial endpoints, and adaptive design changes following interim data analyses can make or break the success of a study.

There are several key methods in clinical trial design that can be enhanced with AI and ML that we will review below:

1. Predictive Modelling to Assist the Selection of the Patient Population

One of the most significant challenges in clinical trial design is to identify the patient population to include in a trial. AI-driven predictive modelling on large datasets, including electronic health records (EHRs), genetic information, and other biomarkers, are now being used to identify patients who are most likely to benefit from a treatment. For example, the FDA recently utilised machine learning techniques to identify a suitable patient population for an emergency use authorisation of Anakinra for COVID-19 Treatment. During the SAVEMORE trial, AI was used to predict which COVID-19 patients would have high suPAR levels, a biomarker associated with severe disease outcomes. By combining predictive modelling with clinical trial data, the FDA was able to identify patients who would most likely benefit from Anakinra treatment under an Emergency Use Authorisation (EUA)^[1]. This method marked a significant milestone in how AI can be leveraged for clinical trial design to make a regulatory decision.

1. Natural Language Processing (NLP) for Hypothesis Generation

The use of NLP in analysing unstructured data, such as patient notes or scientific literature, is becoming more important in clinical trials. NLP algorithms can sift through vast amounts of data to identify patterns, extract relevant information which can be instrumental in generating clinical trial hypotheses.

For example, Novartis utilised NLP in its "data42" project^[2] to digitise and analyse over 20 years of clinical trial protocols and patient data. This enabled the identification of previously unknown correlations between drugs and diseases, facilitating the generation of new hypotheses for clinical trials. By converting unstructured historical data into structured formats, NLP helps researchers explore connections that would be difficult to discover manually, ultimately informing more targeted and efficient clinical trials.

3. Advanced Retrospective Analysis: Leveraging Clinical Results to Detect Heterogeneous Treatment Effects

In clinical trials, especially those with complex datasets, the ability to retrospectively analyse results is crucial for understanding how different subpopulations respond to treatment. Advanced retrospective analysis involves using sophisticated statistical methods and AI-driven tools to detect heterogeneous treatment effects, identify subpopulations that respond differently, and evaluate endpoints likely to succeed in future trials.

Companies like PhaseV specialise in applying these advanced analytical techniques to past trial data. By leveraging AI and machine learning, these analyses can identify specific subpopulations within a trial cohort who may respond more favourably to the treatment.

This approach is particularly valuable in personalised medicine, where treatments may only be effective in specific genetic or biomarker-defined subgroups. By understanding these nuances through retrospective analysis, trial designs can be refined to increase the likelihood of success in future studies. The identification of relevant endpoints, informed by previous trials, also ensures that future trials are designed with a higher probability of demonstrating efficacy.

Learn more about Quanticate's Partnership with Phase V 

4. Adaptive Trial Designs

Adaptive trial designs allow modifications to trial parameters based on interim data. ML plays a crucial role here by providing real-time data analysis and suggesting necessary adjustments to trial protocols. This can include altering dosage, changing patient groups, or modifying endpoints. The FDA has increasingly supported adaptive designs, recognising their potential to improve the efficiency and ethicality of clinical trials^[3].

Quanticate brings extensive expertise in biostatistics and clinical trial design, making them an ideal partner for supporting your adaptive design trial needs. Moreover, Quanticate has developed a sophisticated system of macros that allows for the automated production of complex outputs enabling real-time analysis of interim results crucial for enabling adaptive modifications during the trial.

Challenges and Future Directions

Despite the potential, the use of AI and ML in clinical trials is not without challenges. These include the need for high-quality data, the risk of algorithmic bias, and the regulatory hurdles associated with the adoption of new technologies, particularly around the transparency and explaining how AI models work without compromising patient safety or privacy. However, the future looks promising, with continued advancements in AI leading to more personalised and efficient trial designs.

As AI and ML technologies continue to evolve, their role in clinical trial design is likely to expand. This will not only accelerate the drug development process but also lead to more precise and patient-centred approaches to medicine.

The ongoing innovations in adaptive trial designs, the use of AI for patient selection and retrospective analysis, as well as hypothesis generation demonstrate the industry's commitment to improving clinical trial design. These advancements are essential for addressing the growing complexity of medical research and ensuring that new therapies reach patients as quickly and safely as possible.

The FDA’s support for these technologies, as evidenced by its guidance and real-world applications, underscores the importance of integrating AI and ML into clinical trial designs. As regulatory bodies, industry stakeholders, and technology providers continue to collaborate, we can expect further breakthroughs that will reshape the landscape of clinical trial design.

Leverage AI and ML to Enhance Your Clinical Trial Design

At Quanticate, we are at the forefront of integrating advanced AI and ML techniques into clinical trial design. Our team of experienced statistical consultants offer cutting-edge solutions to improve patient selection, enhance hypothesis generation, and implement adaptive designs, all while ensuring compliance with regulatory standards. Whether you're looking to streamline your trial design, conduct advanced retrospective analyses, or incorporate real-time data for adaptive trials, Quanticate is here to support you every step of the way.

For more information on how we can enhance your clinical trial with AI and ML, please contact Quanticate or submit an RFI today.

Reference:

1. Using Machine Learning to Identify a Suitable Patient Population for Anakinra for the Treatment of COVID-19 Under the Emergency Use Authorization | FDA
2. NLP Analyzes the Past to Inform the Future of Clinical Trial Design (appliedclinicaltrialsonline.com)
3. Adaptive Design Clinical Trials for Drugs and Biologics Guidance for Industry | FDA