The next evolution of digital twins in clinical trials will focus on:

• Wider regulatory acceptance – More agencies incorporating digital twins into trial frameworks
• More advanced AI-driven modeling – Increasing accuracy and predictive power
• Expanded real-world applications – Applying digital twins beyond trials, including in real-world evidence studies
• Earlier-stage drug development – Using digital twins to inform trial design and preclinical research

As computational power and AI capabilities continue to evolve, digital twins will become an even more powerful tool for accelerating clinical development and bringing life-saving treatments to patients faster.

Digital twins are fully reproducible, meaning their results can be independently verified. The process for generating each twin is predefined in the statistical analysis plan, including:

• Model weights, code, and parameters to ensure consistency
• No post-treatment updates—DTGs are not modified with data collected during the trial to prevent bias

While Unlearn continuously refines its models outside of trials, digital twins used in clinical studies follow strict pre-specified protocols, ensuring unbiased, precise treatment effect estimation.

Regulators are recognizing the potential of digital twins.

• The European Medicines Agency (EMA) has qualified PROCOVA™, a digital twin-powered approach for Phase 2 and 3 trials with continuous outcomes.
• The FDA concurs with the EMA, stating that PROCOVA aligns with existing regulatory guidance.

By taking a rigorous, pre-specified approach to digital twins and engaging with regulators early, Unlearn is helping pave the way for broader adoption of AI-driven trial methods.

A digital twin is a computational model of a real-world entity designed to simulate and predict how that entity will change under different conditions.

Within a healthcare context, digital twins are virtual representations (“digital twin”) of patients that are generated from multimodal patient data and population data. Unlike other computational models, a digital twin is specifically tied to an individual patient and is used to simulate and forecast future health outcomes under different scenarios.

In clinical trials, Unlearn’s digital twins are AI-generated models of individual trial participants. These models use historical patient-level data and advanced machine learning to generate comprehensive forecasts of clinical outcomes, helping optimize clinical development from trial design through execution.

1. Smaller Control Groups, More Patients on Treatment

Randomization is the gold standard in clinical trials, necessitating large control groups. AI-powered digital twins generate forecasts of each participant’s control outcomes, enabling smaller control groups while maintaining trial integrity. This allows more participants to receive the investigational treatment, improving both the efficiency and ethical balance of the trial.‍

2. More Precise and Powerful Studies

Digital twins are trained on vast datasets, capturing disease progression patterns and individual patient characteristics. By reducing variability and improving statistical power, they enhance the ability to detect a meaningful treatment effect—even with fewer participants. This enables more confident decision-making sooner, optimizing trial efficiency and resource use.

3. Accelerating Trial Timelines

By improving patient selection, optimizing study designs, and reducing control group size, trials can reach meaningful results faster. This means fewer delays, quicker decision-making, and potentially faster regulatory approvals.

4. Reducing Costs Without Sacrificing Rigor

Recruiting and monitoring control group patients is costly. By optimizing control group sizes, digital twins can help improve trial efficiency without compromising statistical rigor. More efficient trials mean faster insights and better resource allocation, allowing companies to make more informed decisions about future studies and potential therapies.

5. Better Patient Representation and Inclusion

Traditional trials often struggle to recruit diverse, representative patient populations. By optimizing control group sizes, digital twins allow more patients to be assigned to the investigational treatment arm, which may help improve representation in trials. While digital twins themselves do not create new patient diversity, they support trial designs that maximize the inclusion of underrepresented groups in active treatment arms.

No. Digital twins do not replace actual patients in clinical trials. Instead, they complement traditional trials by generating forecasts of individual patient health outcomes under placebo or standard-of-care conditions. This approach enables smaller, more efficient control groups while maintaining rigorous randomization, improving trial design and resource allocation.

Digital Twin Generators (DTGs) create digital twins using individual participant data from past clinical trials and observational studies. These datasets include:

• Longitudinal clinical records from over 1 million participants
• More than 30 disease indications
• Detailed patient data: demographics, clinical outcomes, labs, biomarkers

Each dataset is carefully curated to match target patient groups and trial requirements. To ensure high-quality data, Unlearn applies rigorous quality control measures—filtering out outliers, duplicates, and low-quality data points—so digital twins remain accurate and reliable.