Clinical trial design simulator utilising real world data

TrialKey is an essential tool for optimising clinical trial designs using AI. By analysing over 350,000 trials, our advanced TrialGen module simulates trials, recommends protocols, and predicts success probabilities with over 90% accuracy – a market-leading capability. It is versatile and can be used for any trial and any phase, including novel trials.

Clinical trials are the foundation of medical progress, yet traditional methods have remained unchanged for over 30 years, making them slow, expensive, and lack transparency. TrialKey leverages predictive analytics and machine learning to enhance efficiency, reduce costs, and accelerate the time to market for new drugs and medical devices.

In employing AI and GPT-4 to analyse real-world trial data, we are able to provide users with:

How does TrialKey work?


TrialKey’s datasets undergo a rigorous curation and categorisation process based on variables, meticulously crafted to guarantee the credibility of the results we deliver to our users.

Data Collection

TrialKey gathers public data from various trial registries such as and ANZCT.

Data Processing

The collected data is curated and processed to create features. This involves a combination of structured data and features generated by Natural Language Processing (NLP) models.

Variable Categorisation

The variables derived from the data can be categorized into two main types:

Trial execution variables

These include details about the trial’s execution such as patient numbers, protocol specifics, inclusion/exclusion criteria, primary/secondary endpoints, principal investigators, Clinical Research Organisations (CROs), and trial sites.

Technology variables

These encompass information related to the drug being studied, such as its mechanism of action, administration method, dosage, and frequency.

Target Definition

The target of the TrialKey algorithm is to predict whether a trial will successfully meet its primary endpoint within the corresponding phase.

Machine Learning Modeling

TrialKey employs various advanced machine learning models to optimise prediction accuracy. These models are experimented with and fine-tuned to maximise their ability to predict trial outcomes accurately.

In this holdout dataset:

➜ 982 trials were actual positives

➜ The top 10% of algorithm predictions identified 42.5% of actual positives with over 90% precision

➜ Extending to the top 20% predictions, more than 80% of trial successes were captured with precision consistently exceeding 90%

This methodology, combined with the utilisation of a model trained on a wide range of variables, provides robust insights into the algorithm’s performance from 2017 to 2022.

As a testament to its reliability, TrialKey accurately forecasted the success of the Moderna and Pfizer vaccines during the COVID-19 pandemic before their clinical trials concluded. This success underscores TrialKey’s potential to empower patients, healthcare professionals, clinical researchers, and investors with the foresight needed for effective decision-making.

Join us in our mission to revolutionise clinical trials, accelerating and optimising drug development to improve medical outcomes for patients globally.

Optimise Clinical Trials with AI-Driven Real World Data Insights

TrialKey stands as the market leader for clinical trial prediction and design optimisation, with an accuracy rate of +90%