run better and avoid injuries. AI makes it possible.

As a runner, I pay as much attention to tracking my runs as the amount of wear and tear my running shoes endure. This is important because continuing to run in shoes with damaged soles can lead to serious injuries.

While completing the “Intro to AI Product Design” certificate from ELVTR in 2024, I set out to solve this problem, which I find very relevant to me as I’m sure it is to many other long-distance runners. I solved it with the knowledge and understanding I acquired from this course of how artificial intelligence actually works.


A “RUNNING” PROBLEM

Runners depend on the quality and state of their running shoes not only for performance, but also to prevent injuries, such as shin splints, ankle sprains and “runner’s knees”, amongst others.

After certain number of miles, the inside of running shoes sole is no longer able to provide the necessary support it was designed to offer runners (despite the fact the shoe may still look in good shape).

Medically-verified data indicates that running on overused shoes increases the chances of injuries.

Today, runners have to manually and proactively keep up with running shoes mileage. This is done without regard to other difficult to track factors that may affect shoe deterioration such as weather, type of terrain or other details about how the shoes are built.

Other fitness apps like Strava or RunKeeper offer very basic shoe tracking features, it at all. CENTRUN was created to offer a solution to this unaddressed problem.

CENTRUN is the result of an effort that showcases how designers like myself can effectively use the AI tools at our disposal today to create a useful product or service.
 

What is “centrun”?

CENTRUN is app for runners that goes beyond simple run tracking. It has an AI-powered feature that allows users manage the lifespan of all their running shoes by pairing their run data with information about the weather and terrain the shoes are being used on, as well a information of how each shoe is built in order to predict when it will be time to replace the shoes.

The app goes a step further by recommending the best replacement shoes. Using the power of AI, CENTRUN takes into account data from runners with similar running characteristics who have used similar shoes, as well as data from shoe manufacturers to accurately recommend the best shoes to every runner.

CENTRUN even offers the ability to shop for recommendations right in the app.

 

Visual Design process & strategy

To start, I used Wireframe Designer (a Figma plugin) by feeding it an initial idea of how I thought the home screen of the app should look like. I used the following prompt:

Create the home view of an app called "CENTRUN". This an app that is meant for runners. This home view has 3 main sections: "Home" "Shoes" and "Training". These 3 sections are located at the bottom inside a menu bar that goes from side to side. This home view shows a dashboard that displays the current running shoe a user is running in with the amount of miles ran in that shoe. It also shows a list of the last few runs the user has gone in. It also displays a countdown for the next half marathon coming up in 4 weeks.

This was the result:

I then used Conjure AI (another Figma plugin)to get a general look and feel of what the app could look kike using a very similar prompt. These were some of the initial results:

I ultimately chose this as the basis for the visual direction:

Then I kept using Conjure AI to come up with the logo for CENTRUN. I specified in the prompt the logo needed to represent running, and that it needed to be green and grey.

I ultimately chose this one:

Lastly, I tried QoQo.AI for a couple of things. I tried an option called “design review” which, upon feeding it a similar prompt to the one I used in Wireframe Designer, it ended up showing me a whole strategy depicting how I could actually go about making this app a reality.

QoQo.AI was also a great tool for the steps of understanding and analysis in a typical design thinking process. I used a feature called “design brief”, with (again) a similar prompt to the first one I used. It hen generated a design brief with valuable information I could use to further develop my idea (in case this was a real product I were to be building)

I used these AI-powered tools mainly for inspiration and not to depend on them for the actual final product. They saved me time in ways I could not have imagined before –especially when it came to another important step in the design thinking process: ideation (particularly with Wireframe Designer and Conjure AI)

 

The role of ai behind the scenes

AI wasn’t only used for information architecture and visual design. Machine Learning is what powers the innovative solution of the “Shoe Advisor“ feature in CENTRUN.

CENTRUN was designed to generate key data that, when paired with other ambient data and shoe construction features, can easily help runners know when to replace their shoes and what to replace them with.

In this case, AI acts as an advisor. It presents various shoes alternatives that are adequate to each runner. It also instructs users on the various characteristics present in different shoes, which allows them to make informed decisions on future purchases that can benefit their performance.

CENTRUN accurately predicts when it is time to change a shoe based on data such as user input (tracking runs; indicating which shoe was used for each run); as well as ambient data gathered by the app (e.g. weather conditions at the location during the time of the run, or the type of terrain & elevation, which affect the type of deterioration a running shoe experiences). The app also gathers data directly from the users with ratings and opinions on the shoes they are currently using. The incentive for users to provide this type of voluntary data is to receive more accurate recommendations based on actual user reviews other runners may be contributing to CENTRUN). 

Lastly, runners’ bio-metrics are collected (weight, height, foot shape, even previous injuries, amongst others) and paired with shoe data (from CENTRUN’s database with information provided by manufacturers such as shoe material, heel drop, cushioning and other materials in the shoe).

Machine learning algorithms used in CENTRUN’s main value prop are:

  • Linear regression, which provides information on the relationship between an independent variable (in this case, the user’s running) and a dependent variable (in this case, the shoe’s deterioration based on how it’s built). 

  • Time series analysis looks at a timeline of events or data points to understand patterns or trends over time to predict when the shoes might need to be replaced based on how much they are being used. This, paired with other data (like a predetermined training plan) help with prediction accuracy.

  • Clustering allows CENTRUN to sort data into different groups based on their similarities (e.g. similar types of runners in terms of their running goals, bio-metrics, shoe preferences and shoe usage patterns) to generate more personalized and accurate predictions.

 

AI MEETS INTERACTION DESIGN

Once the AI portion of the app was determined, it was time to connect my interaction design skills to the machine learning algorithms that power CENTRUN.

I created a series of task flows to determine the main flows a user would experience throughout the app:

Replacing a shoe based on an AI-powered recommendation (Task flow).

Data & information flow details.

These task flows and their corresponding details were then used (as they typically would in any design project) to create the high-fidelity mockups in Figma. They were naturally based on all the initial visual design explorations previously shown, for which AI-powered tools were also instrumental.


Conclusion

CENTRUN was a great opportunity for me as a user experience designer to put into actual practice all the knowledge I gathered on how AI works from my time learning from the ELVTR course materials and assignments.

The impact that knowing how to apply AI had on this project was greatly felt in two main areas:

  • Concept strategy

  • Design thinking process

Having a solid understanding of how AI, machine learning and deep learning allowed me to pair the problem I set out to solve with solutions that AI is greatly position to solve —primarily by using various source of data to feed the right algorithms. Applying this level of knowledge in AI to a real-world problem was incredibly empowering and inspiring.

Lastly, it was evident to me how impactful AI can be in terms of the tools that can be used all throughout the typical design thinking process. From understanding, research and analysis, all the way through to the ideation, prototyping and testing stages, AI-based tools enabled me to work faster and more effectively than ever before.

Having realized the impact of AI on my practice as a UX designer is what ultimately inspired me to share these learning with other designers who could feel as empowered as I did. I partnered with RE:THINK to create a seminar-style talk called “Practical Methods for Integrating AI Into Your Design Practice”, which I’ve had the pleasure of giving at various points throughout 2024.