Patent Application
20210177329
Uroflowmetric tests generally measure the flow of urine. Uroflowmetry may track many aspects of a patient voiding, such as tracking how fast urine flows, how much urine flows out, and how long it takes a patient to fully void. By measuring the average and top rates of urine flow, uroflowmetry tests can show various health concerns regarding your urinary tract. However, there is always a need to provide more accurate data to a physician to better diagnosis health concerns.
Some aspects of the disclosure are directed toward a method for providing uroflowmeter data. The method may comprise receiving volume sample data representative of volume sample data from a uroflowmeter device, calculating the slope of the volume sample data, and performing additional actions if the calculated slope reaches a trigger threshold. If the calculated slope reaches a trigger threshold, the method may further include the steps of determining if an artifact is present in the volume sample data. This may include comparing the morphology of the potential artifact to morphologies of known artifacts and comparing the value of the volume sample data before and after the potential artifact. If an artifact is determined to be present in the volume sample data and the volume sample data before the potential artifact is less than or equal to the volume sample data after the potential artifact, remove the portion of the volume sample data which represents the detected artifact.
In some embodiments, calculating the slope of the volume sample data may comprise calculating an average slope and adjusting the slope by the average slope. Furthermore, the average slope may be calculated using a least-squares best fit model. In some embodiments, the trigger threshold comprises a lower slope value. For example, the trigger threshold may be when the slope is at or below 0 mL/s. An artifact may comprise an event, such as at least one of a door opening, a door closing, an HVAC system running, footsteps, and mechanical vibrations.
In some embodiments, receiving volume sample data comprises receiving a plurality of sample data points. In such embodiments, receiving the plurality of volume sample data points may further comprise receiving the volume sample data points from a buffer, such as a rolling buffer.
In some embodiments, determining an artifact is present may comprise determining a baseline, wherein the baseline is an area prior to the start of the potential artifact; determining a post-baseline, wherein the post-baseline is an area after the end of the potential artifact; determining a trough, wherein the trough is the lowest local minima bounded by the trigger and time after the event; determining a post peak, wherein the post-peak is the largest local maxima bounded by the trigger and the post-baseline; determining an onset. Determining the onset may comprise at least one of: locating a local minima or flat area in the span prior from the pre-peak that is less than 10% of the pre-peak amplitude from the baseline, wherein the located local minima or flat area is the onset, or locating the point with steepest positive slope between the baseline and the trigger, then locating point the flattest point between the point with the steepest positive slope and the baseline, wherein the located point is the onset. Furthermore, in some embodiments, if the pre-peak is greater than the baseline and the post peak, the artifact is identified as a positive form artifact; and if the trough is after the trigger, the artifact is identified as a negative form artifact. Additionally or alternatively, determining a baseline delta, wherein the baseline delta is the difference between the baseline and the post-baseline; and if the baseline delta is above a certain value, then the potential artifact is determined to not be an artifact. In such embodiments, the method may further include the steps of determining a peak-to-trough amplitude, wherein the peak-to-trough amplitude is the difference between the lowest and highest value between the baseline and the post-baseline; and if the baseline delta value is higher than 15% of the peak-tough amplitude, determine that the potential artifact is not an artifact.
As discussed herein, the uroflowmeter 110 may comprise uroflowmeter sensor 113. Uroflowmeter sensor 113 may comprise a variety of sensor types, such as a load cell, a transducer, or the like. In such examples, the external events 145 may momentarily cause a higher or lower pressure applied to the uroflowmeter sensor 113, affecting the pressure inside the uroflowmeter pressing down on the uroflowmeter sensor 113. The difference in pressure may then result in an artifact being present in the data.
As discussed herein, a uroflowmeter may comprise a sensor (e.g. a load cell, a transducer, or the like). In such examples, the external events may momentarily cause a high or lower pressure applied to the uroflowmeter sensor, affecting the pressure inside the uroflowmeter pressing down on the sensor. The difference in pressure may then result an artifact being present in the data.
Furthermore, in some embodiments the uroflowmeter 110 may be covered or partially covered by a silicon boot or other material to make the system water-proof or more water-resistant. In such examples, there may be a localized pressure difference between the inside and outside of the silicon boot. Accordingly, when external pressures are present the silicon boot may act as a diaphragm which results in additional force being applied to a pressure measurement device (e.g. uroflowmeter sensor 113). Accordingly, the uroflowmeter can have an increased likelihood of detecting pressure waves resulting from an external event 145 which can appear as an artifact or noise in the data.
The external events 145 may come from a variety of sources. For example, an external event 145 may be from a door opening/closing, an HVAC system running, footsteps, mechanical vibrations (e.g. from heavy machinery), movement of the container 120 or portions of the uroflowmeter, or other situations which may emit substantial vibrations and/or noises.
In some embodiments, the data gathered by the uroflowmeter 110 is filtered, such as low pass filtered or bandpass filtered, to reduce and/or eliminate noise from environmental and/or external conditions. In such examples, a 5 Hz lowpass filter can be used because uroflowmeter signals related to physiological sources (e.g. voiding) generally have frequencies of less than 5 Hz, or in some examples less than 1.5 Hz. In some embodiments, the target band can be about 0-1 Hz, 0-1.5 Hz, 0.4-0.8 Hz, or other target bands known to one of ordinary skill in the art. However in some examples, some noise artifacts may have a frequency of less than 5 Hz, such as artifacts from external events 145, and therefore cannot be simply eliminated by lowpass or bandpass filtering without also filtering out important physiological information.
Furthermore, as more data is being gathered from the uroflowmeter 110, such as by collecting from an increased bandwidth collection, the uroflowmeter 110 can collect more external events 145 in the data. Such artifacts in the data may cause the data to include non-uroflowmetry information, be harder to read, and potentially lead to a user (e.g. a physician) interpreting inaccurate results.
To overcome such a problem, a noise artifact detection method may be used as a technical artifact detector which may operate on volume channel data from uroflowmeter 110. The noise artifact detection method can be used to identify localized atmospheric pressure artifacts, vibration artifacts, or the like related to external events 145 near a sensitive measuring device (e.g. uroflowmeter 110) such as the opening and closing of doors. Additionally, the noise artifact detection method as discussed herein may be performed using a noise artifact detection system, such as a system incorporated into uroflowmetry system 100 as shown in
The noise artifact detection method may be performed in the uroflowmeter 110. Additionally or alternatively the noise artifact detection method may be performed on a separate device, such as an external computation device 150 (e.g. a computer, tablet, smartphone, or the like). In such embodiments, the uroflowmeter 110 may be in communication with the external computation device 150, such as via connection 155. Connection 155 may comprise a variety of connections know in the art, such as a wired connection, a wireless connection, a combination thereof, or the like.
The noise artifact detection method may be used to analyze the morphology of affected waveform shapes in the data produced by uroflowmeter sensor 113. More specifically, the noise artifact detection method may be used to identify artifacts from various external events 145 in the data. In some examples, voiding patterns will not be identified as artifacts. However, other external events, such as taping, splashing, footsteps on the floor, or the like may be identified as artifacts based on various parameters, such as amplitude, frequency, and/or shape of the signal detected by the uroflowmeter 110. When a portion of the data is identified as an artifact, said portion of the data may be removed, interpolated, marked, or the like. For example, a FlowNoiseDataRetraction Event may be initiated and/or generated for the timespan of the artifact, as discussed herein.
As discussed herein, various artifacts may be identified. In some embodiments, artifacts may come in distinct forms.
A positive form artifact as shown in
Additionally, a check 325 for a trigger value may be performed, wherein if no trigger is found (e.g. NO in check 325) then the method reverts back to the initial search step 320 and if a trigger is found (e.g. YES n check 325) then the data collection step 330 may be initialized. In some embodiments, an initial trigger may start again immediately. However, in other embodiments, an initial rigger may happen periodically, or based on other factors (e.g. user input, flow rate, other sensor information, etc.). After data collection in step 330 is initialized, a check 335 to see if the needed amount of samples has been collected. In some embodiments, the samples may be collected at a rate of 100 samples a second, however other sampling rates, such as rates above or below 100 samples a second have been contemplated. In some embodiments, the data collected in step 330 may be placed in a buffer, such as a circular buffer or the like. The needed amount of samples can be an amount of samples, samples spanning an amount of time, or the like. In some embodiments, the samples needed may be based on a predetermined amount. After the needed samples have been collected (e.g. YES in check 335) the analysis step 340 may be performed. Then in some embodiments, another initial search may be performed after analysis is complete. In some embodiments, an initial search may be performed simultaneously with analysis. Various other embodiments similar to those described with respect to
In some embodiments, samples (e.g. volume samples from uroflowmeter sensor 113) are put through a bandpass filter after being received and before being analyzed such as by method 300. Additionally or alternatively, the bandpass filtering may be performed at various other times, such as during the priming step 310, the initial search 320 and/or during the data collection step 330. In some embodiments, every data sample used in during the analysis step 340 may be filtered (e.g. bandpass filtered) prior to the analysis step 340. The bandpass filter may be used to remove frequencies outside of the major power center of a targeted artifact as well as partially normalize the data with respect to concurrent flow. Various bandpass filters may be used, such as ones with various orders as well as ranges. In some examples, a 64th order bandpass filter with a finite impulse response (FIR) of 0.4 Hz to 0.8 Hz may be used. In some embodiments, the normalizing the data may comprise normalizing the amplitude against an estimate flow pattern based on the volume data before and after the area under consideration. In some instances, this estimation may be based on a fit line, however other models may be used such as higher order polynomials, exponential models, logarithmic models, or the like.
After samples go through the bandpass filter, depending on which step shown in method 300 is currently being performed, the processing may continue with the processed values. For example, if the method is currently performing the data collection step 330, the process may continue to the analysis step 340.
With respect to
In some embodiments, the initial search step 320 may comprise a method which determines the various trends of data received from a uroflowmeter (e.g. volume data). This may comprise an algorithm to calculate a current slope in the data. In such embodiments, the slope may be calculated using a best fit model, such as a least-squares best fit model.
In some embodiments, if the calculated slope is above or below a trigger threshold (e.g. YES in check 325), the method may record the collected sample, such as by a sample stamp. In some embodiments, the trigger threshold may be when the calculated slope is at or below 0 mL/s. When the calculated slope falls below 0 mL/s artifact detection can be initiated because the overall volume should not decrease during a uroflowmetry test or the like. Additionally, the trigger threshold may be other values, such as values above or below 0 mL/s, such as −1.2 mL/s. In some embodiments, the threshold may be a value at or less than 0 mL/s. Furthermore, trigger thresholds may be set based on the patient, such as age gender, weight, medical conditions, or other conditions known to one of ordinary skill in the art. In some embodiments, the threshold detection is after an initial lowpass and/or bandpass filter is used to reduce the number of false artifacts.
When the trigger threshold is met (e.g. YES in check 325), the method may proceed to the data collection step 330. In some embodiments, the data collection step may comprise collecting all necessary samples prior to performing the analysis step 340. In some examples, data can be collected for a given amount of time, such as 15000 ms after a trigger threshold is found, such as starting from the sample stamp as discussed herein. However, other amounts of time, such as more than 15000 ms and less than 15000 ms have been contemplated. Alternatively, data may be collected until a specified amount of samples have been collected. In some embodiments, data may be continually collected and stored in a buffer, such as a rolling buffer. In such embodiments, the rolling buffer may collect data over a predetermined period of time (e.g. 15000 ms), for a predetermined amount of data samples, or the like.
Once the samples have been collected (e.g. YES in check 335) the method may continue to the analysis step 340. In some embodiments, the analysis step 340 may be performed simultaneously to the data collection step 330 and/or the samples may be stored for later use.
As described herein, the analysis step may be performed to determine whether or not an artifact is present after an event has occurred (e.g. a trigger threshold has been reached). In some embodiments, the analysis step 340 may comprise determining whether or not the morphology of the potential artifact represents or closely represents the morphology of a known artifact, such as a positive form artifact (e.g. opening a door) and a negative form artifact (e.g. closing a door) as shown in
With respect to
In some examples, the best fit may be subtracted from the volume data in order to provide a more relatively flat dataset as shown in
Turning back to
In some embodiments, method 300 may look for a particular event prior to determining whether or not an artifact complex is present (e.g. performing analysis step 340). As described herein, the particular event may be the flow rate going below a threshold value, such as 0 mL/s, −1.2 mL/s, or the like.
During the analysis step, the method may look for a few based elements to help determine which form the artifact is in as well as the start and/or end of the artifact complex. The basic elements may comprise one or more of an initial baseline, trough, post-peak, pre-peak, and post-baseline which are described in further detail herein and shown with respect to
An initial Baseline can be defined as the area prior to the start of the artifact. In some embodiments, the initial base line is calculated as an arithmetic mean (e.g. average) of a window, or grouping, of the process sample values before the trigger. Alternatively, the initial baseline may use other methods to calculate its value, such as a best fit line from least means squared, an R2 value, or the like.
A trough can be defined as the lowest local minima in the window, or group of samples, bounded by the trigger and a post-baseline. In no local minima can be determined or found, then the trough can be defined as the most-flat time bounded by the trigger and the post-baseline.
A post-peak can be defined as the largest local maxima in the window, or group of samples, bounded by the trigger and the post-baseline.
A pre-peak can be defined as the largest local maxima in the window, or group of samples, bounded by the initial baseline and the trigger.
In some embodiments, an event (e.g. positive form, negative form) may need to have an identified trough to be considered as an artifact. In some examples, the positive form artifact may be identified if the pre-peak is greater than the initial baseline and the post-peak. Additionally or alternatively, the negative form artifact may be identified if the identified trough is after the trigger. As described herein,
If the form is believed to be a positive form, additional elements may be determined. In some examples, the onset can be determined. The onset can be defined as the point where the positive deflection begins prior to the peak (shown in
For example, first method for determining the onset may comprise locating a local minima or flat area in the span prior from the pre-peak that is less than 10% of the pre-peak amplitude from the initial baseline. With respect to
In some examples, the artifact complex may be rejected as a potential artifact if no onset is identified.
After the onset is identified, the duration of the artifact can be estimated. In some examples, the duration of the artifact may be estimated using the following empirical ratio:
Duration Artifact=Amplitude Peak-to-Peak*40 EQ. 1:
Wherein the Duration Artifact is the duration of the artifact from the onset in milliseconds (ms) and the AmplitudePeak-to-Peak which can be the difference between the peak and the trough within the potential artifact, or the difference between the pre-peak and the trough and can be measured in milliliters (mL). The value of 40 may be based on the used uroflowmeter (e.g. uroflowmeter 110) and the sensor used (e.g. uroflowmeter sensor 113). In some embodiments, a value of above or below 40 may be used to calculate the Duration Artifact depending on the system and/or the surrounding environment.
The estimated duration (e.g. Duration Artifact) calculated using EQ. 1 can be bounded by a maximum estimated duration, as discussed herein. Additionally, an endpoint can be checked against any detected elements; and if any parts of the artifact complex are found outside of the estimated range, that point is set as the endpoint.
Once the bounds are calculated, a normalized peak-to-peak amplitude can be calculated for the entire artifact complex and checked to see if it meets a threshold value. The normalized peak-to-peak amplitude may be calculated based on the normalization of the initial baseline or an additional normalization. In some embodiments, the data may be normalized based on a line between the start and end of the potential artifact (e.g. a line between the initial baseline value and the post-baseline value. In some embodiments, the threshold value may be 0.30 mL, however values above 0.30 mL and below 0.30 mL have been contemplated. Additionally or alternatively, the peak-to-peak amplitude may be based on the data between start and end of the potential artifact.
In some embodiments, potential artifacts can be evaluated to see if the peak positive flow is above a threshold. In such embodiments, the threshold may be between 0.4 mL/s and 0.6 mL/s, such as 0.48 mL/s; however values above 0.6 mL/s and below 0.4 mL/s have been contemplated.
After the artifact complex is identified (e.g. as a positive form, negative form, not an artifact, etc.). A holdoff point may be used and then the method may switch back into the initial search step 320. In some embodiments, the holdoff point may be 10 ms, however values above 10 ms and below 10 ms have been contemplated.
If the form is believed to be a negative form, additional elements may be determined. In some examples, the onset and post-baseline may be determined. The onset can be defined as the point where the negative deflection begins in the window prior to the trigger. In some examples, this can be found by looking for the earliest point of negative slope in the window. The post-baseline can be defined as the flattest slope in the area after the post-peak. In some examples, the artifact complex may be rejected as a potential artifact if no onset and post-baseline are identified.
After the onset and the post-baseline are identified, the duration of the artifact can be estimated. In some examples, the duration of the artifact may be estimated using the following empirical ratio:
Duration Artifact=Amplitude Peak-to-Peak*40 EQ. 2:
Wherein Duration Artifact is the duration of the artifact from the onset in milliseconds (ms) and the Amplitude Peak-to-Peak, which can be the difference between the trough and the post-peak or the trough and post-baseline and can be measured in milliliters (mL). In some embodiments, a value of above or below 40 may be used to calculate the Duration Artifact depending on the system and/or the surrounding environment.
As discussed herein, the estimated duration (Duration Artifact) calculated in equation 2 can be bounded by a maximum estimated duration. Additionally, an endpoint can be checked against any detected elements; and if any parts of the artifact complex are found outside of the estimated range, that point is set as the endpoint.
Once the bounds are calculated, a normalized peak-to-peak amplitude can be calculated for the entire artifact complex and checked to see if it meets a threshold value. The normalized peak-to-peak amplitude may be calculated based on the normalization of the initial baseline or an additional normalization. In some embodiments, the data may be normalized based on a line between the start and end of the potential artifact (e.g. a line between the initial baseline value and the post-baseline value. In some embodiments, artifact complex's which do not meet the threshold value are not identified as artifacts. In some embodiments, the threshold value may be 0.08 mL, however values above 0.08 mL and below 0.08 mL have been contemplated.
Additionally or alternatively, potential artifacts can be evaluated to see if the peak positive flow is above a threshold. In such embodiments, the threshold may be between 0.4 mL/s and 0.6 mL/s, such as 0.48 mL/s; however values above 0.6 mL/s and below 0.4 mL/s have been contemplated.
After the artifact complex is identified (e.g. as a positive form, negative form, not an artifact, etc.). A holdoff point may be used and then the algorithm may switch back into the initial search state. In some embodiments, the holdoff point may be 10 ms, however values above 10 ms and below 10 ms have been contemplated.
In some embodiments, leak detection is evaluated if the initial baseline and the post-baseline slopes are flat or sufficiently flat, such as shown in
Methods may also include determining and/or calculating a mean value of the baseline areas (e.g. initial-baseline and post-baseline) and may determine a baseline delta based on the differences between the initial baseline and the post-baseline.
In some embodiments, a leak is determined if the baseline delta is greater than a threshold of the normalized Peak-to-Peak amplitude (e.g. Peak-to-Trough amplitude and/or Trough-to-Post-Peak amplitude). In such examples, if the delta baseline is above a certain value, the event is determined to be a leak rather than an artifact from an external event (e.g. positive form, negative form, or the like). For example, the threshold may be 15% of the Peak-to-Trough amplitude and/or Trough-to-Post-Peak amplitude, however other values higher than 15% and lower than 15% have been contemplated. In some embodiments, when the overall volume does not change substantially, or in relation to the size of the artifact, the artifact may be determined to be an artifact from an external event (e.g. positive form, negative form, or the like) rather than a physiological event (e.g., leak, initialization of voiding). Accordingly, in some embodiments a comparison is made between the volume before and after the event and the change in volume is compared to, for instance, the size of the event to determine if the event is something other than a physiological event. Such analysis may be used to trigger the analysis of such an event to determine if a noise artifact is present. Noise artifacts in the data may be identified via many different methods, such as those disclosed herein, as well as other types of known signal analysis and comparisons to atlases of known artifacts and via many types of analyses including via training artificial intelligence to recognize many different noise artifacts.
Additionally or alternatively, if the baseline delta is negative (e.g. the initial baseline is less than the post-baseline) it may be determined that an error has occurred, the sensor (e.g. uroflowmeter sensor 113) is mis-calibrated, or the like. However, in some situations, the baseline delta may have a negative value, such as if the container (e.g. container 120) is bumped and a portion of the liquid within the container is spilled. In such examples, the baseline delta may reflect the loss of liquid within the container. In some embodiments, the post-baseline value may be checked with respect to the trough value and/or the local minimum between the initial baseline and the post-baseline. In such embodiments, if the post-baseline value is less than the trough value and/or minimum value it may be determined that an error has occurred, the sensor is mis-calibrated, or the like.
In some embodiments as discussed herein, there is a holdoff period between trigger point searches (e.g. check 325). However, there is a possibility that a suitable waveform morphology can be found that has an onset within the window of a prior detected artifact complex. In such examples, the subsequent artifact complex may have the onset adjusted such that it occurs only after the holdoff window.
Additionally or alternatively, a leak may be determined based on whether the event fits into one or more predetermined cases. For example, first case may be identifying physiological leaks with a very low flow, a second case may be identifying physiological leaks which are quick/short, and a third case may be identifying artifacts during a flow which may look similar to a quick/short physiological leak.
With respect to the first case, in some situations a physiological leak may be very low, or small however it may still be important to count as a physiological leak rather than an artifact. Identifying such physiological leaks may comprise determining if the flow rate before and after an event is less than a threshold flow rate (e.g. less than or equal to 0.1 mL/s or like). Additionally, identifying such physiological leaks may comprise determining if the post-baseline value goes up by more than a threshold percent of the peak value (e.g. 40% or the like). And if the threshold percent is reached, the event may be determined as a physiological leak rather than an artifact.
With respect to the second case, in some situations a physiological leak may not last long (e.g. be relatively short or quick), however it may still be important to identify as a physiological leak rather than an artifact. Identifying such physiological leaks may comprise determining if the flow rate before and after an event is above a threshold flow rate (e.g. being greater than 0.8 mL/s or the like). Additionally, identifying such physiological leaks may comprise comparing the peak value to the post-baseline value, and if the post-baseline value goes up by less than a threshold percent of the peak value (e.g. 50% or the like), the event may be identified as a physiological leak during flow rather than an artifact.
With respect to the third case, in some situations various artifacts may look very similar to physiological leaks as discussed above with respect to the second case, however it may still be important to count as an artifact rather than a genuine physiological leak. In such situations, to identifying such artifacts may comprise determining if the flow rate before and after an event is below a threshold flow rate, which in some examples may be complementary to the threshold flow rate discussed herein with respect to the third case (e.g. being less than 0.8 mL/s). Additionally, the peak value may be compared to the post-baseline value, and if the post-baseline value goes up by more than a threshold percent of the peak value, the event may be identified as an artifact. In some embodiments, the threshold percent may be complementary to the threshold percent discussed herein with respect to the second case (e.g. greater than 50%).
In some embodiments, a maximum artifact duration may be defined. For example, the maximum artifact duration may be 15000 ms, however durations less than 15000 ms or more than 15000 ms have been contemplated. For example, the maximum artifact duration may be adjusted for a variety of qualities, such as sensor type, uroflowmeter type, location, temperature, air pressure. In some embodiments, as the maximum artifact duration is scaled, other parameters may be additionally scaled by similar amounts, such as the times between the Onset, Pre-Peak, Trigger, Trough, Post-Peak, and Post-Baseline of
In some embodiments, the maximum duration of an artifact detection area is bounded. In such embodiments, two factors may bound the maximum duration. The first factor may be limits on the detection range for morphological elements and the second factor may be the estimated artifact duration based on peak-to-peak amplitude.
In some embodiments, the estimated duration for positive form artifacts is explicitly bounded by the maximum estimated duration for a positive form artifact (e.g. 15000 ms). Additionally, the maximum estimated duration may be applied to artifacts larger than a threshold amplitude. In some examples, the threshold may be calculated using EQ. 1 wherein Duration Artifact is 15000 ms. However other thresholds may be used.
In some embodiments, the estimated duration for negative form artifacts is explicitly bounded by the maximum estimated duration for a positive form artifact (e.g. 12692 ms). Additionally, the maximum estimated duration may be applied to artifacts larger than a threshold amplitude. In some examples, the threshold may be calculated using EQ. 2 wherein Duration Artifact is 12692 ms. However other thresholds may be used.
In some embodiments, when an artifact (e.g. positive form artifact, negative form artifact, or the like) is found, they are omitted from the data sample set. In such examples, the data within the artifact may be interpolated using the data from each side of the artifact. Alternatively, the artifact may be simply marked in such a way to notify a user (e.g. a physician) that the data in that window of time is an artifact rather than diagnostic information.
Additionally or alternatively, an aggressive Urocap Noise Detection (AUND) method may be used as an aggressive noise detector. The AUND method may also operate on the volume channel of a uroflowmeter (e.g. uroflowmeter 110) similar to the noise artifact detection method 300 described herein. In some embodiments, the AUND method can be used to remove a wider range of artifacts. In such embodiments, the AUND method may be used during times where no leak is present, such as times other than near active flows or leaks. However, the AUND method may be used at other times, such as near active flow or leaks. In some embodiments, the AUND method may be used separately from the noise artifact detection method as described herein. Additionally or alternatively, the AUND method may be used tandem. Furthermore, the AUND method as discussed herein may be performing using an AUND system, such as a system incorporated into the uroflowmetry system 100 as shown in
The AUND method may be used to detect period of time, or groupings of consecutively collected sample volume data, called baselines. A baseline, as described herein, may be an interval where there is high confidence in the estimated expected volume value, such as when the uroflowmeter is in a steady state. In some embodiments, if there is an interval between two baselines, and wherein the respective baseline volume values are close enough together such that no flow or leak of significance could have occurred all value changes within the interval between the two baselines can be considered noise. When a portion of the data is identified as noise, said portion of the data may be removed, interpolated, marked, or the like. For example, a FlowNoiseDataRetraction Event can then be generated on that interval to remove all noise and artifacts within. Additionally or alternatively, the data within the interval may be interpolated using any method known to one of ordinary skill in the art.
In some embodiments, the AUND method may consider any intervals within a threshold and any baseline differences within a threshold. In some embodiments, the threshold for the intervals is an interval which is 30 seconds or less; however other intervals may be used, such as more than 30 seconds or less than 30 seconds. Alternatively, a high and low threshold for the intervals may be used, such as intervals between 1 second and 30 seconds. Similarly, the baseline threshold may be values less than 0.4 mL; however other values may be used, such as more than 0.4 mL or less than 0.4 mL. Alternatively, a high and low threshold may be used for the baseline difference, such as values between 0.1 and 0.4 mL.
In some embodiments, the AUND method may be used in real time, such as while sample data is being collected or shortly after (e.g. 1/10th of a second, a predetermined amount of samples, as soon as data is gathered in the buffer, as soon as a threshold is detected, etc.). Alternatively, the AUND method may happen after data collection, such as after a predetermine amount of time after a threshold, upon user input, or after all samples have been collected.
Various algorithms and/or methods may be used to determine baselines within the sample data and the confidence for those baselines. Two exemplary types of baselines which can be used are temporary baselines and global baselines.
Temporary baselines can be used to remove clearly defined artifacts with low latency as opposed to waiting for other methods which may depend on additional samples to define a baseline. For example, if an unstable period in volume sample data is detected, temporary baselines may be used to determine whether or not the unstable period contains a flow, leak, or is an artifact.
Global baselines can be calculated using larger sample intervals when compared to the temporary baselines. As a result, global baselines can have a higher confidence in estimating the baseline values, which may allow for better analysis of whether or not the unstable period in the volume sample contains a flow, leak, or is an artifact. Furthermore, by using more samples, it may be possible to use additional checks to determine if a baseline exists or not.
The mean of an interval can be calculated representing the line of best fit that can be plotted using the samples with the restriction of having a slope of 0. The mean squared error may measure how close the samples within an interval are with respect to the mean. Lower mean squared error values can indicate a higher confidence of the mean value representing the sample interval.
Calculating the R2 value may give an idea of how much correlation exists within the sample interval. In some embodiments, a high R2 value may indicate that the volume samples within the interval are trending up or down with a high degree of confidence. In embodiments comprising a high R2 value, a proper baseline may not be able to be established as the uroflowmeter may not be in a steady state. In contrast, if the R2 value is low, then a proper baseline may be able to be established, as the uroflowmeter may be in a steady state.
In embodiments having more samples, it may also be possible to perform more rigorous calculations when determining the potential baseline value. Rather than using the mean of the sample interval as the baseline value and the mean squared average as the confidence, it may also be possible to use the cluster based median of the sample interval. In such embodiments, samples within the interval may be grouped into value clusters limited by a predefined value for the range of values allowed in each cluster. Then, the baseline reference may be given by the median of samples within the cluster that contains the largest number of volume samples. The confidence of this value may be measured by the percentage of samples in the largest cluster versus the total number of samples analyzed within the interval. As such, a high cluster size percentage can indicate a higher confidence.
A threshold value may be configured for combinations of ambiguous regions between each of the baselines found, as described herein. For example, between two global baselines, between two temporary baselines, between a global baseline and a temporary baseline, or between any two other baselines known to one of ordinary skill in the art. Then as discussed herein, if the change between the baseline values is larger than the threshold value, then the interval is may be considered to be a leak or flow; otherwise, if the change between the baseline values is smaller than the predetermined threshold value, the interval may be considered an artifact and the data comprised within the interval may be removed, interpolated, marked, or the like. For example, a FlowNoiseDataRetraction artifact can be generated or any other interpolation method known to one of ordinary skill in the art.
Various embodiments have been described. Such examples are non-limiting, and do not define or limit the scope of the invention in any way.
This application claims the benefit of US Provisional Application No. 62/948,804, filed Dec. 16, 2019, the contents of which are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 62948804 | Dec 2019 | US |
