skinID: Democratizing the Formation of Skin Cancer Datasets

Diagnosis of skin lesions is often left solely to practicing dermatologists. With the advent of machine learning research in skin cancer pre-diagnosis, our application seeks to diagnose lesions in parallel to with clinicians to provide the most accurate response, minimizing costs associated with follow-up testing to the end-user and providers. In the process, data submitted can be opt-in to contribute to building datasets for underrepresented groups in medical datasets; in the prevalent HAM10000 dataset, a majority to all are pictured patients are light-skinned.