Personalized VR Therapy for Patients with Dementia

Using AI and VR to Improve Reminiscence Therapy for Patients with Dementia

Amsal Gilani
Feb 1, 2019 · 6 min read

Why do patients with dementia have to suffer from anxiety?

Why do they have to go through trauma?

Why do 152 million people need to go through this? (by 2050)

And more importantly — why do the families of these 152 million people have to go through this.

These are the questions that we need to ask ourselves.


During a hackathon, I attended at The Knowledge Society, my team (Azim Jiwani, Ayleen Farnood) and I came up with the idea of using personalized VR reminiscence therapy as a way to treat dementia. After this hackathon, Ayleen Farnood and I decided to take this idea and try to make it a reality. So, we decided to submit our idea to the Google AI for Social Good Grant to potentially receive up to $2M of funding. Let me break down our idea for you:

There are three main problems with the status quo:

According to the World Health Organization, 50 million people in this world suffer from dementia, and in 2050 this figure is anticipated to increase to 152 million.

This is a big deal and a growing problem in our society.

We might not be able to prevent dementia, but what we can do is use exponential technologies to find a treatment for the people with it.

Dementia is an overall term that describes a group of symptoms associated with a decline in memory or other thinking skills severe enough to reduce a person’s ability to perform everyday activities. Unfortunately, it has numerous side effects such as trauma, depression and general discomfort within the lives of the patient and his/her family. Even triggering a single memory of a patient’s life can drastically improve their mental health and wellbeing. The fact that not much progress has been made to help improve current treatment struck our team as an opportunity to do something about it.

Currently, one of the most common treatments for dementia is reminiscence therapy, which is the process of showing common objects from a person’s life to trigger memories. According to studies done, however, the majority of participants in this type of therapy believe that it isn’t personalized enough and often skims over the small details that are necessary to trigger memories.

Only 34% of people actually found that reminiscence therapy had a lasting impact on their memory recall.


Going back to the questions:

Why do patients with dementia have to suffer from anxiety?

Why do they have to go through trauma?

Why do 152 million people need to go through this? (by 2050)

And more importantly — why do the families of these 152 million people have to go through this.

These are the questions that we need to ask ourselves.

And the answer?

They don’t. Not in the next 5 years. In the next 5 years, we will be using technologies like AI and VR to solve this problem.


In fact, that future is almost here — a startup called Virtue created VR experiences of famous events and locations in the past as a new method of reminiscence therapy and saw that close to 80% of its users experienced cognitive improvement.

Wait…so isn’t this being done already? Yeah — but our idea is much more effective.

Our solution expects to bring that number closer to 100% because we want to use personal videos, with the help of AI, to allow for a much more individualistic and personally tailored experience that will be more effective in triggering memory recall.

We want to intersect the technologies of artificial intelligence and virtual reality to create an immersive VR experience of the patient’s past videos. This will be accomplished through the use of AI algorithms to convert 2D videos from phones into 3D objects in order to create a VR environment. Patients will be able to use an app where they can download their personal videos to experience again in virtual reality with the help of a Google Cardboard or Daydream headset.

There are three main processes to our solution:

AI will speed up this process outlined above and it will allow individuals to be able to experience their own, personal videos within a short period of time. AI models such as general adversarial networks and recurrent neural networks can realistically train the model to create accurate results. We can use programs like Deep3D and Iris Prospect VR to help with this process as well. Without using AI, we would not be able to create a personalized 3D environment that is crucial for the VR experience in short periods of time.

Yep — data is going to be the primary source driving our solution. In fact, it is the driving force behind anything using AI!

Since we will be using personalized videos and converting them to VR, we will need to train our AI algorithm with lots of everyday moments in one’s life. We would need regular videos taken from people’s phones in order to train the model to be able to convert them from 2D into a 3D environment. These videos should come from different types of phones from different years so that we can use a varied assortment of data for the algorithm. This way it will be compatible with as many different forms of input videos as possible.

Data acquisition will be quite simple as we can easily use public datasets of videos from phones as well as the videos we can find online through social media such as Instagram, Facebook, and Youtube. These will be our primary sources of data as we need an assortment of data in order to ensure that the personal video input is compatible with the AI model which this varied data (videos) it is trained on.

Ayleen Farnood and I came up with three main criteria to evaluate our success:

If this solution is taken out to the world, we can expect major benefits and outcomes.

  1. Effective Treatment: Our solution serves as a new method of reminiscence therapy to reduce memory loss.
  2. Valuable Insight: This project may provide insight as to ways in which we can continue to improve treatment for dementia.
  3. Improve Mental Health: Many patients with dementia also suffer from poor mental health. VR help with both memory loss and improving one’s quality of life.
  4. Accessible Worldwide: The app can be downloaded by people all across the world, making it highly accessible to anyone.

Going back to the problem — this solution can impact millions around the world. If this solution becomes a reality, organizations such as the Alzheimer’s Society of Canada may find incentive in using our product as a method of treatment to replace current types of reminiscence therapy.

It’s time to change the way we approach dementia.


If you enjoyed this article:

  • Give me a clap (or 50)
  • Check out my LinkedIn profile here
  • Check out my personal website here
  • AND take a look at my complete portfolio here

Data Driven Investor

from confusion to clarity, not insanity

Amsal Gilani

Written by

Grade 11 high school student — AI & VR developer + Regen Med researcher! Feel free to visit www.amsalgilani.com to learn more about me!

Data Driven Investor

from confusion to clarity, not insanity

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade