2 ms·
Stupid me thinks that if they can measure data that is even indirectly related to blood glucose with sufficient quantity they can actually infer glucose level w
by iamgopal 10y ago
Stupid me thinks that if they can measure data that is even indirectly related to blood glucose with sufficient quantity they can actually infer glucose level with quite accuracy. I.e. only with measuring heart beat, oxygen and co2 level at the skin, temperature and time of the day, and sufficient data crunching they can actually know the glucose levels.
- BjoernKW 10y agoFor hyperglycaemia at least ketones would be a significant marker. In sufficient concentration they even have a characteristic smell detectable by humans. So, creating a sensor that measures ketone concentration shouldn't be too difficult. Hypoglycaemia is more difficult to diagnose because apart from the actual blood glucose level there's no definite symptom that doesn't occur with other conditions as well (typical symptoms include headaches, dizziness, confusion, high pulse rate).
- jpttsn 10y agoCareful there. In this domain, you need the measurement to be accurate every time, and especially on atypical days: that one time you run a marathon or get stuck in the elevator with no food. I get scared thinking of training a glucose predictor on big data. It could reach treacherously high accuracy scores, by secretly assuming every day is an average day.