AI Algorithm Support: Scanbo introduces non-invasive blood glucose measurement protocols

For diabetics who need to regularly assess their blood glucose levels, the measurement method of pricking a needle in the finger to take blood has caused psychological and physiological pressure on many people. The good news is that an AI startup called Scanbo is working to end this invasive blood glucose measurement program with off-the-shelf data analysis and diagnostic tools.

The prototype device developed by the company combines three-lead electrocardiogram measurement and photoplethymography (PPG) technology.

By feeding a set of algorithms a measurement for 60 seconds, a fairly promising blood glucose assessment reading can be given.

In addition to noninvasive blood glucose monitoring, Scanbo founder and CEO Ashishsh Raichura claims that this method is suitable for measuring blood pressure.

At CES 2022, he not only shared more technical details, but also demonstrated the utility of traditional blood glucose analysis protocols.

The results show that the measurements for the two schemes are 6.2 and 6.3 mmol/L, respectively, which means that the error of the new device is only a few percent.

We used three electrodes to acquire electrocardiogram (ECG) data, plus an additional PPG measurement scheme for 60 seconds.

After obtaining the raw data, it is analyzed with the help of machine learning (ML) convolution (CNN) and deep neural network (DNN).

Combining all this data, the research team came up with three different sets of machine learning algorithms and then looked at the efficiency of their analysis of blood sugar.

Scanbo is currently seeking to commercialize its products, but before that, it will have to obtain approval from the North American market regulatory authorities, including the US FDA.