Headquartered in Boston, USA, the client is a digital medicine leader believing that effective medicine can be engaging and fun. With this belief at their core, the client develops game-changing “digital medicine” to develop cognitive treatment. One of their breakthroughs is the child prescription treatment for Attention Deficit Hyperactive Disorder (ADHD), the first and only authorized for ADHD.
The Challenge
Cognitive impairments in people significantly impact their quality of life, leading to troubled thinking, concentrating, making decisions, and more. To reduce these impairments, the client wanted to develop an innovative digital therapy in the form of a video game. This one-of-a-kind product aimed to improve patients’ sustained attention leading to their overall well-being.
The challenge of building this product was to generate 3D data points using 2D images, including things like the speed of the vehicle in changing environments. Addressing this essential requisite would help enhance the overall patient experience and clinical efficiency, resulting in better outcomes. With our rich domain experience in innovative healthcare technology and product engineering DNA, Persistent was onboarded to support the client in achieving desired business outcomes.
We had to define the train and test for the raw data without labels and mimic the machine processing of the video game backend by looking at the 2D images sequence. The most challenging task was to find 3D data points using the 2D game images.
The Solution
The client used numerous bots to test the video game. These bots were expected to generate real user data. However, the client faced a unique challenge when their bots were only able to work in the backend and generate simulated data rather than real user data. To overcome this hurdle, we used Artificial Intelligence (AI) and Machine Learning (ML) to create a bot on top of the existing bot. This allowed us to create a visually immersive experience that emulates real gameplay, generating valuable and actionable data and not simulated data.
We developed JSON for the frames of the video game that contained all the necessary information from the current frame. To make the bots even more efficient, we leveraged the power of ML, building models that could extract insights from both the images and the JSON data provided by the client. We also built an entire operational reporting backend that seamlessly ingested data from multiple sources and placed it into a centralized data hub, also developed by us. This helped us to create custom reports for functional departments such as marketing and finance.
The Outcome
We successfully generated the 3D data points using 2D images, as the client wanted. One of the significant achievements of this project was the ability to automatically process the game at a speed that enabled us to automate 50,000 hours of bot testing, a task that would have been impossible to accomplish manually. Also, without automation testing, achieving 100s of hours of gameplay was impossible.
We have been developing a slew of products in the cognitive function space and this product, which is a video game, is one of its kind to help people suffering from cognitive disorders. Without the support of Persistent’s technical expertise in Machine Learning and Artificial Intelligence, we would have invested countless hours and efforts in making this product successful.
