Statistics that quantify a person’s predisposition to diseases such as diabetes and cancer can be reverse-engineered to reveal the underlying genetic data, prompting privacy concerns