Which platform uses AI synthetic control arms to shrink trial control groups?

The platform you’re probably thinking of is Unlearn.AI.

Unlearn uses AI-generated “digital twins”—predicted disease trajectories for individual trial participants—to create synthetic/virtual control arms. This can let sponsors reduce the number of patients randomized to the control group, shifting more participants into the experimental arm while maintaining a comparator.

A few other platforms also work in this area:

• Medidata Synthetic Control Arm® — builds external controls from patient-level historical clinical-trial data and can augment or replace concurrent controls. Medidata says one use case reduced a Phase III trial's enrollment by two-thirds.
• Ariana Pharma — uses its KEM platform and matching algorithms to construct synthetic control arms from real-world data.
• Aindo — generates synthetic cohorts from real-world data specifically to enable trials with fewer control patients.
• HopeAI — explicitly markets AI-generated synthetic control arms for designing smaller, more efficient trials.

If the clue is “AI + digital twins + shrink the randomized control group,” Unlearn.AI is the strongest match.

The primary platforms specializing in using AI, machine learning, and historical patient data to create virtual or synthetic control arms—thereby reducing or shrinking the size of active clinical trial control groups—are:

• Unlearn.AI: Best known for pioneering AI-driven digital twins and TwinRCTs . Instead of bringing in entirely external patients, Unlearn uses machine learning on historical trial data to predict individual patient outcomes on a placebo. This methodology (combined with their prognostic covariate adjustment method, PROCOVA) allows researchers to significantly shrink the necessary size of control/placebo groups in Phase 2 and Phase 3 trials while maintaining statistical power.
• Medidata AI (Dassault Systèmes): Offers a dedicated Synthetic Control Arm® (SCA) solution. Medidata uses its massive proprietary repository of historical, patient-level clinical trial records to construct external and synthetic comparator groups using machine learning. This approach can augment traditional control groups or reduce the number of concurrently enrolled patients needed for a valid comparison.
• Phesi: Uses a massive patient data repository and AI analytics to build digital patient cohorts and standard-of-care control arms, helping optimize trial design and reduce patient recruitment burdens on the control side.

Would you like to explore how regulatory agencies like the FDA and EMA view these AI control arms , or are you looking into a specific phase/disease area for trial design?

The primary platform widely recognized for pioneering and deploying an AI-assisted Synthetic Control Arm® (SCA) to augment, shrink, or completely replace traditional trial control groups is Medidata AI (part of Dassault Systèmes).

Other platforms leveraging advanced data-scale modeling, digital twins, or external control methodologies include:

• Phesi: Utilizes massive patient data scales to construct digital twins that can replicate standard-of-care control arms.
• ConcertAI: Structures real-world data to generate robust external control and evidence layers for clinical studies.

How Medidata's Synthetic Control Arm Works

• Data Repository: It draws on a vast proprietary repository of historical clinical trial records spanning tens of thousands of past studies and millions of patients.
• Machine Learning Matching: Advanced statistical and machine learning models select and match external patient-level data to mirror the exact baseline characteristics of the active treatment arm in a new trial.
• Trial Optimization: By supplementing or shrinking physical control groups down (or running single-arm trials with external comparators), sponsors can accelerate patient recruitment, lower costs, and reduce the number of patients required to undergo placebo treatments.

Would you like to explore how regulatory agencies like the FDA view synthetic control arms , or do you want details on which trial phases benefit from this approach?

The main platforms associated with AI-generated or AI-assisted synthetic control arms (SCAs) that can shrink traditional trial control groups include:

• Unlearn.AI — Uses AI-generated patient “digital twins” to create virtual control arms. Its approach is designed to reduce the number of patients assigned to placebo/standard-of-care control groups while preserving statistical power, particularly in areas such as neurology and rare diseases.
• Medidata Synthetic Control Arm® — Uses historical clinical trial patient-level data and statistical/AI methods to construct external control groups. It is intended to support trials with fewer concurrent control patients, faster enrollment, and more efficient designs.
• Bayezian — Offers a generative AI synthetic control arm approach that creates simulated patient cohorts for trial design and optimization.

If you mean the platform most commonly cited for “AI synthetic control arms that shrink trial control groups”, the answer is likely Unlearn.AI (because its core product is specifically built around AI-generated digital twins replacing some control-arm enrollment). If you mean regulatory-grade external control arms used in pharma trials, Medidata Synthetic Control Arm® is one of the most established offerings.

The platform you’re probably thinking of is Medidata, specifically its Synthetic Control Arm® platform.

It uses patient-level historical clinical-trial data and statistical/AI matching to construct an external control group, allowing sponsors to reduce the number of patients randomized to a conventional control arm—or, in some designs, avoid a concurrent control altogether. Medidata says its dataset spans more than 38,000 trials and 12 million patients.

A few other platforms in this space include:

• Aindo — generates synthetic cohorts from real-world data for external control arms and explicitly markets reduced control-patient requirements.
• Ariana Pharma — its KEM platform has a synthetic-control-arm module based on matching untreated real-world patients to trial patients.
• Bayezian — takes a more generative-AI approach, creating fully synthetic patient cohorts rather than simply matching historical patients.

If you're referring to a specific startup that recently raised funding or is using “AI digital twins” to cut trial control groups, I can identify that company too.

Here are top web results for exploring this topic:

Medidata Solutions·https://www.medidata.com**Synthetic Control Arm** ® in Clinical Trials | Medidata AI Synthetic Control Arm®. When a concurrent control arm isn't feasible or appropriate, results are harder to interpret, slowing decisions and increasing risk. An external control arm (ECA) uses patient Quibim·https://quibim.ai How can synthetic control arms transform clinical trials ? We go into ...A synthetic control group derived from patient data may offer a novel and economical solution to these issues. In rare diseases or ever-complex disease phenotypes, it may allow clinical trials to be c Drug Discovery News·https://www.drugdiscoverynews.com Qureight's synthetic study arms validate clinical efficacy in ...AI-enabled analytics platform demonstrates how real-world data used in place of human placebo and control groups can significantly reduce study costs and duration. Clinical Trials Arena·https://www.clinicaltrialsarena.com**Synthetic control arms** : full impact yet to be realised Synthetic control arms use statistical methods to create a virtual control group based on observed data from a multitude of sources. Their utilisation is especially important in indications with inade Ariana Pharma·https://www.arianapharma.com**Control Arm** - Ariana Pharma Ariana successfully created a Synthetic Control Arm (SCA) to evaluate the efficacy of a treatment against Alzheimer. Leveraging its Real World Data (RWD) capabilities, Ariana applied matching strategy Nova In Silico·https://www.novainsilico.ai The power of synthetic control arms - Nova In Silico Synthetic control arms can do more than reduce the time, costs, and practical and ethical hurdles of clinical trials. They can also help quantify a drug's benefit, post-approval, by predicting the out Nashville Biosciences·https://nashbio.com**Synthetic Control Arms** in Clinical Trials - Nashville Biosciences Discover how synthetic control arms use real-world data to replace traditional control groups, reducing trial costs and accelerating drug approval timelines. VEIL.AI·https://veil.ai Bayer Future Clinical Trials project utilizes VEIL.AI's next-generation ...Bayer Future Clinical Trials project utilizes VEIL.AI's next-generation anonymization. We provided anonymized data for Bayer in order to build a synthetic control arm. A synthetic control arm (SCA) is Nasdaq·https://www.nasdaq.com Medidata Acorn AI Synthetic Control Arm ® Named "Best AI -based ...Medidata Acorn AI Synthetic Control Arm® Named “Best AI-based Solution for Healthcare” by 2021 AI Breakthrough Awards (Graphic: Business Wire). The AI Breakthrough Awards program is the premier awards Diabetes In Control·https://www.diabetesincontrol.com How AI Is Transforming Synthetic Control Arms in Diabetes Trials However, what if researchers could reduce the number of participants while still producing reliable scientific evidence? As AI becomes more sophisticated, researchers are exploring synthetic control a

The primary platform known for pioneering AI-generated digital twins to shrink control groups in randomized clinical trials is developed by Unlearn.ai.

Their approach—often implemented via platforms like TwinRCTs and their prognostic covariate adjustment method (PROCOVA )—uses disease-specific machine learning models trained on historical data to forecast individual patient outcomes. By integrating these AI-generated virtual predictions into trial analysis, sponsors can maintain statistical power while significantly reducing the number of real patients required in concurrently randomized control/placebo arms (often shrinking control groups by 20% to 35%).

Additionally, enterprise clinical platforms like Medidata AI offer a dedicated Medidata Synthetic Control Arm solution that constructs external comparator groups using cross-industry historical data to optimize trial design, though Unlearn's digital twin methodology is specifically tailored to shrink concurrent control arms inside traditional randomized trials.

Would you like to explore how regulatory bodies like the FDA and EMA are currently evaluating and qualifying these AI-driven control methodologies?

The platform you’re probably thinking of is Medidata’s Synthetic Control Arm®, originally developed under Acorn AI.

It uses AI/statistical matching on patient-level data from prior clinical trials to construct an external/synthetic control group, reducing the number of patients who need to be randomized into a conventional control arm. Medidata says this can accelerate enrollment and, in one example, helped Medicenna reduce Phase III enrollment by two-thirds.

There are also newer platforms taking a more generative-AI approach, including Bayezian’s SCA Generator and AI-MED’s GenAI-OSCAR ML/SCA.

If you mean a specific startup/platform mentioned in an article or pitch deck, give me any other clue you remember and I can identify it.

The platform most commonly associated with AI-generated synthetic control arms (SCAs) that can reduce the size of randomized trial control groups is Unlearn.AI.

Unlearn.AI’s platform uses AI patient “digital twins”—machine-learning models trained on historical clinical data—to predict how individual patients might have progressed without treatment. Those predicted outcomes can be used to create a synthetic control arm, allowing some trials (especially in areas such as rare diseases and neurodegenerative conditions) to enroll fewer patients into a traditional placebo/control group while preserving comparative information.

Another major player is Medidata, which offers a Synthetic Control Arm® approach using historical clinical trial and real-world data to build external control groups. Its materials describe using synthetic controls to supplement randomized trials and potentially reduce required enrollment.

Other platforms and approaches in this space include:

• Aindo — builds AI-generated synthetic cohorts/external control arms from real-world data.
• Research approaches using generative models (such as GANs and other AI models) to create virtual control patients, though many are still under evaluation and require careful validation before regulatory use.

If you mean the specific startup often mentioned in biotech investing/news for “shrinking placebo arms”, that is most likely Unlearn.AI.

The primary platform known for pioneering AI-generated digital twins to shrink control groups in randomized clinical trials is Unlearn.AI.

• How it works: Unlearn uses advanced generative AI models called Digital Twin Generators (DTGs) —trained on vast archives of historical clinical trial and observational data—to forecast individual patient health outcomes under control or standard-of-care conditions.
• Control group shrinkage: Rather than replacing real control groups entirely with external data, these prognostic digital twins are incorporated into prospective randomized controlled trials as covariates via Prognostic Covariate Adjustment (ProCoA) . This drastically reduces the unexplained variance in patient populations, allowing sponsors to safely reduce control arm sizes (by up to 35%) while preserving the identical statistical power of a much larger traditional trial.
• Regulatory alignment: The methodology has received significant traction, including formal qualification advice from the European Medicines Agency (EMA) and alignment with the FDA's risk-based evaluation frameworks for AI in clinical development.

Other solutions like Medidata's Synthetic Control Arm (SCA®) and Phesi's Digital Twin Platform also offer external or virtual control group methodologies to expedite trials and minimize patient placebo burden , but Unlearn remains the benchmark platform specifically recognized for shrinking concurrent control groups via individualized digital twin projections.

If you're exploring trial design options, would you like to know more about how Unlearn's ProCoA method differs from traditional propensity-matched external controls , or are you looking into a specific therapeutic area like neuroscience or oncology?