Studies approximate that 1 in 5 American adults suffer from a mental illness each year. In addition, nearly half of those who do suffer from mental illnesses don't seek treatment or help. As university students, friends, and family members, we've seen how mental health has the ability to impact social, academic and work lives. Therefore, we wanted to create a tool that leverages existing mental health survey data in order to promote awareness of mental health issues and encourage companies and organizations to invest in the mental well being of their employees.

What it does

Our application is intended for employers and organizations to see how aware their policies and culture is of mental illnesses within the workplace. User would need to fill out a short survey on some characteristics of their company or organization such as the number of employees, location, mental health programs, health benefits and ease to get help. Peace of Mind uses this data and compares it historical mental health survey data in order to assign the user's organization a mental health awareness score. This score would be accompanies with research backed suggestions on what they could implement within the organization to raise the score or what they should continue to pursue in order to make sure that the workplace is a healthy environment for their employees. We wanted to show that companies have a lot to gain, economically and ethically, by having healthy, happy, and productive employees in their offices. Therefore, today employees should be able to count on their employers' support as they face mental health illnesses.

How we built it

We got our data from a public mental health survey that 2014 and 2016. We then cleaned and prepared this data in Microsoft Excel. Then we used Microsoft Azure Machine Learning Studio in order to construct our model to predict how likely a company may be contributing to the mental health illnesses of their employees. Tuning our model involved considerations of which column of data would be important to this final prediction. For example, companies and organizations with mental health benefits and a wellness program that educated employees about mental health services would be more likely to have their employees seek help and recover quicker. We then built a web application using html/css/javascript that acted as an interface to inform visiting users about mental health statistics, calculate a mental health awareness score based on a short survey, and give a little background on the research that supports the score. The application was deployed using Microsoft Azure.

Challenges we ran into

We ran into challenges with integrating our web service model with our web application through http requests. This was due to Microsoft Azure Machine Learning Studio needing a workaround to send http requests by using a python script or wrapping in another API. Therefore, we did not have enough time to create a full flow on the application and work on the interface for the survey entry and about portions of the web page.

Accomplishments that we're proud of

We are proud of getting a useful model built to predict whether an employee would go seek help for a mental illness using pretty sparse mental health survey data. Our model achieved a 0.76 F1 score given less than 2000 data points with a few columns after cleaning. Given that this survey is currently being performed again in 2017, there will be even more data available in the near future to train and improve our model with.

What we learned

This was the first machine learning project that we as individuals and as a team tackled so it was interesting to learn about the work and methods needed to clean and prepare data sets to become usable. We also learning about different machine learning algorithms and how tuning their parameters makes better models to predict our data.

What's next for Peace of Mind

Since we have a few separate parts of our project done, the next step is to integrate all of them into a single user experience that someone would be able to use. We will also be excited to see when the mental health survey data for 2017 comes out so that there is more data to train our model. Overall, we hope that our project achieves its goal of promoting awareness for mental health and hopefully can inspire the investment in programs that can help prevent and treat mental illnesses.

Share this project: