Cumulative biosensor system to detect alcohol
A wearable device may be provided for detecting cumulative alcohol consumption. Such a wearable device may include an adhesive layer that adheres to skin and that allows sweat from the skin to pass through and a customizable ink layer that reacts irreversibly to change color along a gradient as ethanol is detected in the sweat. The customizable ink continues to increase color intensity along the gradient as ethanol continues to be detected in the sweat over time.
1. A method performed by a processor executing instructions stored in memory for detecting cumulative ethanol consumption, the method comprising:
receiving input indicative of a plurality of reactions at a plurality of corresponding sections of a wearable biosensor at a point in time, wherein the reactions occur in response to detection of ethanol consumption;
identifying the plurality of corresponding sections of the wearable biosensor from the received input including comparing the corresponding sections of the wearable biosensor to that of one or more reference shapes stored in a database memory by using a transformation function;
identifying the cumulative amount of ethanol consumption by evaluating each of the plurality of corresponding sections of the wearable biosensor relative to a corresponding reference parameter, wherein each of the corresponding sections of the wearable biosensor is activated at a different ethanol consumption level along a gradient;
storing the cumulative amount of ethanol consumption in memory, wherein the cumulative amount of ethanol consumption is stored in association with the point in time; and
identifying a trend of the cumulative ethanol consumption over time.
2. The method of claim 1 , wherein receiving the input comprises receiving an image of the wearable biosensor captured by a camera after the wearable biosensor visibly indicates the reaction.
3. The method of claim 2 , wherein identifying the plurality of corresponding sections of the wearable biosensor from the received input further includes using a sliding window of a convolutional neural network for image patch labeling.
4. The method of claim 2 , wherein identifying the plurality of corresponding sections of the wearable biosensor from the received input further includes extracting one or more masked image regions that define the corresponding sections of the wearable biosensor.
5. The method of claim 4 , wherein extracting the masked image regions includes detecting keypoints that define a shape of the extracted masked image regions.
6. The method of claim 5 , further comprising defining the shape of the extracted masked image region as a deformation of shape of the one or more reference shapes.
7. The method of claim 6 , wherein comparing the corresponding sections of the wearable biosensor to the one or more reference shapes comprises mapping from the shape of the one or more reference shapes to the shape of the extracted masked image region by the transformation function.
8. The method of claim 2 , further comprising normalizing the image for color to eliminate effects of illumination.
9. The method of claim 2 , wherein evaluating each of the plurality of corresponding sections of the wearable biosensor relative to a corresponding reference parameter comprises:
detecting a color value of at least one corresponding section; and
matching the detected color value to a stored chart of calibrated color concentration values.
10. The method of claim 9 , wherein matching the detected color value to the stored chart includes comparing the detected color value to a representative sample of skin types along a Fitzpatrick phototyping scale.
11. The method of claim 2 , further comprising normalizing the image for color to preserve inherent colors of the image of the wearable biosensor and colors of one or more reference colors stored in a database memory.
12. The method of claim 2 , further comprising normalizing the image for color to preserve color variations associated with the cumulative amount of ethanol consumption.