Risk Management
As with any project, in order for it to be successful, we must analyze the various risks associated with the project and determine how to best reduce their impact on the overall project.
Below, we've outlined a number of what we have identified as the most vital risks, as well as how we plan to minimize their impact on our project.
Security of User's Data and Inputs
Security of user's data always must be of utmost importance to any company, especially when the product involves results of an analysis of the user's mental health. To mitigate the chance of personal information leaks, our product will assure to securely encrypt user data, and dissociate information retrieved directly from the user whenever possible. This primarily will involve only allowing summarized user data to be sent from the device.
Incorrect Prediction of Results
Machine Learning tools rely on the fact that the result of their calculations is only a prediction. There is bound to be error in those predictions. In our product, we will be sure to communicate the accuracy of results to users, as well as refer users to a properly trained mental health professional or resource whenever more accurate results are vital to a user's well-being.
Access to Training Data
In order to teach a machine learning model how to identify symptoms of depression, the machine must first have access to a dataset of inputs from identified users of depression. While we have found a set that provides this kind of data, it requires action by the Stevens Institute of Technology legal team, which has thus far taken longer than expected. In order to avoid impact on the project's timeline, we have collected our own dataset in a manner that we believe to be approximate to the dataset we were attempting to attain. Until the Stevens' legal team completes it's necessary actions, the dataset we have attained will allow us to continue our project timeline on track without delay.