Usable and Reliable Consumer-Facing Mobile Health Applications
There have been increasing number of mobile health applications available for general population. However, the usability and inproper guidance could result in suboptimal performance, if worse, misdiagnosis. To minimize human error in self-diagnosis using mobile phones, I have been building an smartphone application to guide users to perform the task correctly in real-time using computer vision techniques. I am also exploring a method to ensure sensor inputs (e.g., image, audio, accelerometer) are compatible for the machine learning models to perform at its best performance.