Classically, the diagnosis of type 1 diabetes comes after a patient presents with unexplained weight loss, extreme thirst and frequent urination and a lab test reveals off-the-charts blood glucose levels. At the 85th Scientific Sessions of the American Diabetes Association in Chicago, researchers highlighted two options – a blood test and a machine learning model – for diagnosing the disease much earlier in its progression, when damage to the pancreas' beta cells could be slowed.