What is a Digital Twin?

Unlearn is excited that founder and CEO Charles Fisher will be giving a conference session at DPHARM 2019! He'll be sharing how digital twins will revolutionize more powerful and efficient clinical trials.

The topic of his talk brings up an important question: what is a digital twin? 

Broadly speaking, a digital twin is a virtual copy, or simulation, of some real object. The digital twin concept was originally developed in engineering and has been used for many applications, such as building airplanes [1].

Let's go through an aerospace example to illustrate what is required to make a digital twin, how it's created, and how it's used.

The key part of creating a digital twin is developing a model of the real object - in our example, a real jet engine. To build a physics model of the engine, we need to incorporate two elements: the design specifications, which give us parameters for how the engine should work, and real engine data, which tell us how the real engine behaves in particular conditions (fig. 1A).

How does the physics model create a digital twin? We first collect data describing the current state of a real engine and combine it with the engine's specifications to create a physics model. This model is designed to simulate how our particular engine will behave in the future. Since the simulation is our digital twin, the model is able to make digital twin predictions (fig. 1B).

So how is the digital twin used? We can monitor the performance of both the real engine and digital twin simultaneously over time and compare the two. In our graph (fig. 1C), the digital twin predictions show relatively stable performance, whereas the real engine shows variable performance that sometimes dips below the predicted performance. Since the real engine shows less than optimal performance, we can use our comparison to fix any relevant issues with the engine. 

Translating this example to the clinical world, how is using digital twins in aerospace related to digital twins in clinical trials? 

Creating a digital twin - a simulated patient - in clinical trials also requires building a model, but there is one key difference. In the case of our engine, we have design specifications that can be used to build a physics model. But we don't have design specifications for patients. So to build our model, we need to incorporate two slightly different elements: a database of patient data (medical records) and machine learning, which uses these data to simulate disease progression for patients (fig. 2A).

How does our machine learning model create a digital twin? We first train a model to create realistic patient simulations using a database of historical patient data (fig. 2B, part 1). Then, we collect medical records that describe the current state of a particular patient and input the data into our model. This model is designed to simulate how the patient's disease will progress in the future. Again, since the simulation is our digital twin, the model is able to make digital twin predictions (fig. 2B, part 2).

How do we use digital twins in the clinical setting? Like the engine, we can monitor how a disease progresses for an actual patient versus the digital twin over time. In fig. 2C, we are monitoring progression by ADAS-Cog scores, a commonly used metric to evaluate cognitive decline for an Alzheimer's disease patient - the higher the score, the more severe the decline. Let's say we have an Alzheimer's patient who is taking a drug in a clinical trial, and we create a digital twin that acts as a control without the drug. If the actual patient maintains healthy cognitive function i.e. a lower score, compared to the digital twin with unhealthy function i.e. higher score, this result could indicate that the drug being tested is effective and help it advance to the next clinical trial phase. 

This post was designed to be a basic introduction to digital twins. If you are interested in learning more, here are some ways to do so:

  1. Attend Charles Fisher's zeitgeist talk at DPHARM today (Sept. 17) at 4:20pm. 
  2. Follow Unlearn on Twitter or LinkedIn.
  3. Stay tuned for an upcoming blog post that further explores the concept of the Unlearn Digital Twin™ in clinical trials!


1. For more information about digital twin technology in the aerospace industry: https://www.aviationtoday.com/2018/09/14/boeing-ceo-talks-digital-twin-era-aviation/

Enter your email address to download paper.

Click the link to begin download.
Oops! Something went wrong while submitting the form.

Enter your email address to watch the webinar.

Click the link to watch webinar.
Oops! Something went wrong while submitting the form.
White Papers

Using AI-based Prognostic Models to Design Efficient, Unbiased Clinical Trials


Part 3: Innovation in Clinical Research: AI-based Drug Development Tools and the Regulatory Landscape‍


Part 2: Faster, More Efficient Trials: Novel Trial Designs using Digital Twins‍

Watch an overview of specific use cases for Digital Twins and learn how novel trial designs with Digital Twins enable smaller trials that maintain their power.
Watch a panel discussion on the regulatory landscape where experts share perspectives on the future of AI-based drug development tools like Digital Twins.
Statistical principles of clinical trials with Digital Twins