Method of obtaining high accuracy urination information
A method of obtaining high accuracy urination information is proposed. There may be provided the method of obtaining the urination information, wherein sound data is divided into a plurality of windows, segmented target data corresponding to respective windows is obtained from the sound data, segmented classification data classifying urination sections or non-urination sections and segmented urine flow rate data are obtained by using the obtained segmented target data, and urination data is obtained by using the obtained segmented classification data and the segmented urine flow rate data.
1. A method of obtaining urination information, the method comprising:
obtaining sound data by using a sound sensor;
obtaining, by a processor, a first plurality of segmented target data corresponding to a first plurality of windows from the sound data, wherein each of the first plurality of windows includes a first length and is sequentially determined between a starting point and an ending point of the sound data;
obtaining, by the processor, a second plurality of segmented target data corresponding to a second plurality of windows from the sound data, wherein each of the second plurality of windows includes a second length and is sequentially determined between the starting point and the ending point of the sound data;
obtaining, by the processor, a plurality of segmented classification data by sequentially inputting the first plurality of segmented data into a classification model, wherein the classification model is configured to output data comprising at least a value for classifying a urination section or a non-urination section when data related to urination sound are inputted;
obtaining, by the processor, a plurality of segmented urine flow rate data by sequentially inputting the second plurality of segmented data into a prediction model, wherein the prediction model is configured to output data comprising at least a value for urine flow rate when data related to urination sound are inputted; and
obtaining urination data using at least the plurality of segmented classification data and the plurality of segmented urine flow rate data.
2. The method of claim 1 ,
wherein the first plurality of segmented target data includes first to m-th segmented target data and the first plurality of windows includes m windows,
wherein the first to m-th segmented target data corresponds to the m windows,
wherein the m is a natural number greater than or equal to 2,
wherein the second plurality of segmented target data includes first to n-th segmented target data and the second plurality of windows includes n windows,
wherein the first to n-th segmented target data corresponds to the n windows, and
wherein the n is a natural number greater than or equal to 2.
3. The method of claim 2 , wherein consecutive windows among the n windows partially overlap each other.
4. The method of claim 3 ,
wherein consecutive windows among the m windows partially overlap each other, and
wherein an overlapping degree of the consecutive windows among the m windows is different from an overlapping degree of consecutive windows among the n windows.
5. The method of claim 3 ,
wherein consecutive windows among the n windows partially overlap each other, and
wherein an overlapping degree of the consecutive windows among the m windows is same as an overlapping degree of consecutive windows among the n windows.
6. The method of claim 2 , wherein consecutive windows among the m windows do not overlap each other.
7. The method of claim 2 , wherein the each of m windows are same as each of the n windows.
8. The method of claim 2 , wherein the obtaining of the first plurality of segmented target data comprises:
transforming the sound data to spectrogram data; and
obtaining the first to m-th segmented target data corresponding to the m windows from the spectrogram data.
9. The method of claim 2 ,
wherein the obtaining of the first plurality of segmented target data comprises:
obtaining first to m-th segmented sound data corresponding to the m windows, and
transforming each of the first to m-th segmented sound data to spectrogram data to obtain the first to m-th segmented target data.
10. The method of claim 1 , wherein the obtaining of the urination data comprises:
obtaining urination classification data using the plurality of segmented classification data;
obtaining candidate urine flow rate data using the plurality of segmented urine flow rate data; and
processing the candidate urine flow rate data using the urination classification data.
11. The method of claim 10 , wherein the urination data are obtained through convolution operating of the urination classification data and the candidate urine flow rate data.
12. The method of claim 10 , wherein the urination data are obtained through multiple operating of at least part of the urination classification data and the candidate urine flow rate data.
13. The method of claim 1 , further comprising:
correcting the urination data by using a compensation value.
14. The method of claim 13 , wherein the compensation value is obtained by:
obtaining sample sound data including sound of urination,
obtaining a predictive value by using the sample sound data, the classification model, and the prediction model, wherein the predictive value represents voiding volume,
obtaining a measured value by measuring voiding volume corresponding to the sample sound data, and
obtaining the compensation value by using at least the predictive value and the measured value.
15. A non-transitory computer-readable recording medium having recorded thereon one or more computer readable instructions which, when executed by at least one processor of an electronic device, cause the electronic device to perform operations comprising:
obtain sound data recorded by a sound sensor;
obtain, by the at least one processor, a first plurality of segmented target data corresponding to a first plurality of windows from the sound data, wherein each of the first plurality of windows has a first length and is sequentially determined between a starting point and an ending point of the sound data;
obtain, by the at least one processor, a second plurality of segmented target data corresponding to a second plurality of windows from the sound data, wherein each of the second plurality of windows has a second length and is sequentially determined between the starting point and the ending point of the sound data;
obtain, by the at least one processor, a plurality of segmented classification data by sequentially inputting the first plurality of segmented data into a classification model, wherein the classification model is configured to output data comprising at least a value for classifying a urination section or a non-urination section when data related to urination sound are inputted;
obtain, by the at least one processor, a plurality of segmented urine flow rate data by sequentially inputting the second plurality of segmented data into a prediction model, wherein the prediction model is configured to output data comprising at least a value for urine flow rate when data related to urination sound are inputted; and
obtain urination data using at least the plurality of segmented classification data and the plurality of segmented urine flow rate data.
16. The method of claim 1 , wherein the first length is same as the second length.
17. The method of claim 2 , wherein the first length is different from the second length.