GB2624434A

GB2624434A - Methods relating to the anaerobic threshold

Info

Publication number: GB2624434A
Application number: GB2217271.2A
Authority: GB
Inventors: Krishan Santokhi Jay
Original assignee: Dyson Technology Ltd
Current assignee: Dyson Technology Ltd
Priority date: 2022-11-18
Filing date: 2022-11-18
Publication date: 2024-05-22
Also published as: GB202217271D0; WO2024105592A1

Abstract

A user’s aerobic-to-anaerobic threshold is determined using a wearable air purification apparatus 100 (e.g. headphones) during physical exercise. The wearable device has a pressure sensor 152 (e.g. in a visor 120) to detect pressure changes caused by the user’s breathing. Breathing rate over time is monitored. A breathing value indicative of the user’s aerobic-to-anaerobic threshold is identified based on a change in the user’s breathing rate over time. The alteration in breathing rate may be indicated by an inflection point on a breathing rate plot. The user may be warned when they exceed or fall below the threshold in order to target exercise intensity, and measurement over time of anaerobic threshold may be used to measure progress/improvement. AI is used to identify non-breathing related events in order to de-noise the signal.

Description

Intellectual Property Office Application No G1322172712 RTM Date May 2023 The following terms are registered trade marks and should be read as such wherever they occur in this document: Bluetooth Intellectual Property Office is an operating name of the Patent Office www.gov.uk/ipo

METHODS RELATING TO THE ANAEROBIC THRESHOLD

Field of the Invention

The present invention relates to methods relating to the anaerobic threshold, which may also be known as the aerobic-to-anaerobic threshold or the lactate threshold.

Background

The aerobic-to-anaerobic threshold (below 'anaerobic threshold') is a physiological breakpoint at which the human body's metabolism transitions from aerobic to anaerobic. More particularly, the anaerobic threshold corresponds to the lowest intensity of physical exercise where production of lactate begins to exceed the body's capacity to clear lactate and, at the same time, the highest intensity of physical exercise that can be maintained for prolonged periods. Knowledge about a person's anaerobic threshold is therefore considered to be a relevant factor for targeted physical exercise for improving of fitness.

Since the anaerobic threshold is individual to a person, it is necessary to measure or estimate the anaerobic threshold. Moreover, the anaerobic threshold changes as a person's fitness changes and it may therefore be desirable to repeatedly measure this threshold or subsequently adjust an initial estimate to take into account a change in fitness.

Traditionally measurements of the anaerobic threshold involve gas exchange measurements and blood sampling procedures in a laboratory for the assessment of the maximal exercise intensity above which blood lactate rises inexorably and cannot be stabilised. While this provides an accurate assessment, the procedure is expensive and may be unaffordable for many athletes and sports coaches.

As an alternative to laboratory assessment, a rule-of-thumb estimate for reaching anaerobic threshold may be given as 80% of a maximum heart rate, where the maximum heart rate is roughly estimated as 220 minus age in years. However, such estimates may be very inaccurate, and may not be sensitive to a person's fitness, meaning that specifically targefing exercise around the estimated anaerobic threshold may not yield the desired results.

The present invention has been devised in light of the above considerations.

Summary of the Invention

As an alternative to the laboratory-based measurement and the heartrate-based estimation of the anaerobic threshold, here a method is described that can yield an adequately accurate estimate without requiring a laboratory.

According to a first aspect of the invention, there is provided a method of determining a user's aerobic-to-anaerobic threshold using a wearable air purification apparatus during physical exercise. The method uses a pressure sensor of the wearable air purification apparatus configured to detect pressure changes caused by the user's breathing. The method includes generating a first dataset by detecting output from the pressure sensor over a first time period; analysing the first dataset to monitor a user's breathing rate over the first time period; determining a first breathing rate value indicative of the user's aerobic-toanaerobic threshold based on a change in the user's breathing rate over the first time period.

By using a pressure sensor interacting directly with local pressure changes caused by user breaths, a more accurate determination of the user's aerobic-to-anaerobic threshold may be obtained than by inferring the breathing rate from other physiological signals (e.g. heart rate variability from photoplethysmography) more susceptible to error, or other means for approximating the aerobic-toanaerobic threshold (e.g. heartrate).

In some examples, determining the breathing rate value indicative of the user's aerobic-to-anaerobic threshold may include identifying an elbow point in the user's breathing rate over the first time period. An elbow point can be understood to be a local maximum of curvature.

In some examples, the method may further include notifying the user of the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.

In some examples, the method may further include detecting output from the pressure sensor to monitor a user's breathing rate over a second time period, which takes place later than the first time period; determining that the user's breathing rate approaches the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold; notifying the user (e.g. to adjust their level of exercise) if the user's breathing rate approaches the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.

For purposes of targeted physical exercise, the user may want to keep exercise intensity either below the aerobic-to-anaerobic threshold (for stamina training) or above the aerobic-to-anaerobic threshold (for strength training). By monitoring the user's breathing rate, it is possible to detect whether the user's breathing rate approaches the breathing rate value indicative of the user's aerobic-to-anaerobic threshold. If so, the user can be notified about exceeding or, as the case may be, falling below the user's aerobic-to-anaerobic threshold. The determination that the user's breathing rate approaches the first breathing rate value may involve projecting the user's breathing rate, or may involve the user's breathing rate reaching within a predetermined threshold of the first breathing rate.

In some examples, the method may further include generating a third dataset by detecting output from the pressure sensor over a third time period, which takes place later than the first time period; analysing the third dataset to monitor a user's breathing rate over the third time period; determining a second breathing rate value indicative of the user's aerobic-to-anaerobic threshold based on a change in the user's breathing rate over the third time period.

By determining a second breathing rate value indicative of the aerobic-to-anaerobic threshold at a later time, a shift of the aerobic-to-anaerobic threshold may be identified which may occur, for example. as a result of regular physical exercise and improved physical fitness.

In some examples, the method may further include notifying the user of the second breathing rate value indicative of the user's aerobic-to-anaerobic threshold.

In some examples, the method may further include applying a low-pass filter to the first dataset By applying the low-pass filter to the dataset or datasets recorded by the pressure sensor, high-frequency signals present therein may be reduced to ease identification of the user's breathing at a comparatively low frequency.

In some examples, the method may further include analysing the first dataset to monitor the user's breathing rate over the first time period includes using a peak finding algorithm to determine the number of breaths over the first time period.

By using a suitable peak finding algorithm, which are generally known algorithms, breaths may be identified as peaks in in the dataset.

In some examples, analysing the first dataset to monitor user's breathing rate over the first time period may include: determining whether a subset of the first dataset is indicative of a non-breathing event using a trained machine-learned model that has been trained with labelled datasets; and excluding the subset indicative of a non-breathing event when determining the first breathing rate value.

By identifying and excluding non-breathing events, such as talking, coughing or sneezing, the accuracy of the method may be improved.

In some examples, determining the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold may comprise: transmitting the first dataset or data representative thereof to a remote server for the server to determine the first breathing rate value; and receiving from the remote server the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.

By using a remote server for determining the breathing rate value indicative of the user's aerobic-to-anaerobic threshold, greater computational power may be utilised.

In some examples, determining the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold may comprise transmitting the first dataset or data representative thereof to a smart device in communication with the wearable air purification apparatus for the smart device to determine the first breathing rate value.

By using a smart device for determining the breathing rate value indicative of the user's aerobic-toanaerobic threshold, greater computational power may be utilised.

In some examples, the method may further include receiving from the smart device the first breathing rate value indicative of the users aerobic-to-anaerobic threshold.

According to another aspect of the invention, there may be provided a method of training a machine learning model for use in a method according to as described above, wherein the method includes: generating multiple reference datasets, wherein each reference dataset is respectively generated by monitoring output over a time period from a pressure sensor configured to detect pressure changes caused by a user's breathing; producing labelled training data by labelling portions of the reference datasets indicative of non-breathing events to associate those portions with non-breathing events; using the labelled training data to train a machine learning model for identifying a non-breathing event in the first dataset.

According to another aspect of the invention, there may be provided a method of digitally coaching a user relative to the user's aerobic-to-anaerobic threshold during physical exercise, the method including: retrieving a breathing rate value indicative of the user's aerobic-to-anaerobic threshold; detecting output from a pressure sensor of a wearable air purification apparatus to monitor a user's breathing rate; notifying the user if the user's breathing rate approaches the breathing value indicative of the user's aerobic-to-anaerobic threshold.

By monitoring the user's breathing rate relative to a breathing rate value indicative of the user's aerobic-to-anaerobic threshold, it is possible to inform a user when approaching said threshold. Thus, the user may adjust exercise intensity, for example in order to ensure training below said threshold or training above said threshold.

According to another aspect of the invention, there may be provided a wearable air purification apparatus configured to carry out the method as described above, comprising: a fan assembly configured to provide a filtered airflow; a visor to direct the filtered airflow to a user who is wearing the air purification apparatus; a pressure sensor to detect the user's breathing.

In some examples, the apparatus is configured to transmit the first dataset or data representative thereof to a remote server or a smart device.

The invention includes the combination of the aspects and preferred features described except where such a combination is clearly impermissible or expressly avoided.

Summary of the Figures

Embodiments and experiments illustrating the principles of the invention will now be discussed with reference to the accompanying figures in which: Figure 1 illustrates a wearable air purification apparatus.

Figure 2 illustrates onboard components of the wearable air purification apparatus.

Figure 3 shows the wearable air purification apparatus, a smart device and a server.

Figure 4 illustrates a method of determining the anaerobic threshold.

Figure 5 is a plot of sensor data generated by a pressure sensor of the apparatus of Figure 1.

Figure 6 is a low-pass-filtered plot of the sensor data of Figure 5.

Figure 7 is another low-pass-filtered plot of the sensor data of Figure 5 and illustrates a peak detection algorithm applied to the low-pass-filtered signal where each peak is indicated with a 'x' and corresponds to a single breath.

Figure 8 is graph showing a breathing rate over time.

Figure 9 shows a trendline generated based on the data of Figure 8.

Figure 10 shows the trendline of Figure 9 and marks a breathing rate indicative of the anaerobic threshold.

Figure 11 illustrates a method of training a machine learning model. Figure 12 illustrates a method of digital coaching.

Detailed Description of the Invention

Aspects and embodiments of the present invention will now be discussed with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art.

Figures 1 and 2 illustrate a wearable air purification apparatus 100 embodying aspects of the present disclosure. Figure 1 shows the apparatus 100, while Figure 2 is a schematic illustration of onboard components of the apparatus 100.

The wearable air purification apparatus 100 is configured to be worn on a wearer's head and deliver a filtered airflow towards a lower nasal and mouth region of a wearer's face. Consequently, the wearer's exposure to ambient air pollution may desirably be reduced.

The wearable air purification apparatus 100 comprises a headgear 110 for mounting on a wearer's head and generating filtered airflows, and a visor 120 (or 'nozzle') for directing the filtered airflows from the headgear 110 towards the lower nasal and mouth region of the wearer's face. In this example, the headgear 110 has the form of "over-the-ear" headphones comprising a head band 130, and left and right housings 140, 150 connected to respective ends of the headband 130. The headband 130 is generally elongate and arcuate in form, and is configured to overlie a top and sides of the wearer's head in use.

The headband 130 is formed to resiliently hold the housings 140, 150 against opposite sides, i.e., left and rights sides respectively, of the wearer's head, to thereby firmly retain the apparatus 100 mounted on the wearer's head such that the visor 120 extends across the wearer's face, approximately directly in front of the lower nasal and mouth region.

The left and right housings 140, 150 are arranged to house components, including loudspeaker assemblies 141, 151 for generating sound, and fan assemblies 142, 152 for generating filtered airflows.

Suitably, each fan assembly 142, 152 comprises an air inlet, an air outlet, a filter and a fan driven by a motor to draw in ambient air through the air inlet and to discharge filtered airflow from the air outlet. Such arrangements of fan assemblies are known and detailed description thereof is therefore omitted. The visor 120 directs the filtered airflows discharged from the fan assemblies 140, 150 towards the lower nasal and mouth region of the wearer's face.

The air purification apparatus 100 includes an air pressure sensor 152, a memory 154, a communicator 156 and a controller 158. In this example, the pressure sensor 152 is included in the visor 120 while the memory 154, the communicator 156 and the controller 158 are provided in the housings 140, 150.

The pressure sensor 152 is configured to detect pressure changes caused by the user's breathing.

Suitably, the pressure sensor 152 is arranged to directly interact with local pressure changes caused by user breaths. The pressure sensor 152 transmits an output signal indicative of the wearer's breathing.

Utilising data collection by the pressure sensor 152 of the head-worn apparatus 100, it is possible to obtain the breathing rate directly rather than inferring the breathing from other biological signals (e.g. heart rate variability from photoplethysmography) which may be more susceptible to error than the described approach.

The memory 154 is configured to store a dataset corresponding to the output signal from the pressure sensor 152, i.e. the stored dataset is indicative of the wearer's breathing.

The communicator 156 is configured to communicate with remote devices, for example through Bluetooth.

The controller 158 is configured to control the pressure sensor 152, the memory 154 and the communicator 156.

Figure 3 shows the air purification apparatus 100, a smart device 200, and a server 300. In this example, the air purification apparatus 100 is communicably coupled to the smart device 200 by means of the communicator 156 using a suitable wireless connection, e.g. through Bluetooth or W-Fi. The air purification apparatus 100 and the smart device 200 are configured to exchange data through this connection. As such, output from the pressure sensor 152 can be transmitted from the air purification apparatus 100 to the smart device 200. Similarly, data can be transmitted from the smart device 200 to the air purification apparatus 100, for example to control the air purification apparatus 100 to issue a notification to the wearer.

The smart device 200 is communicably coupled to the server 300 by means of a suitable wireless connection, e.g. a cellular connection or Wi-Fi, to exchange data.

In use, sensor data is recorded by the onboard pressure sensor 152 during physical exercise, transmitted to the smart device 200, and then transmitted by the smart device 200 to the sever 300 for analysis. The results of the analysis, including the determined anaerobic threshold, are transmitted by the sever 300 to the smart device 200, where the user of the smart device 200 may access this information.

Figure 4 illustrates a method of determining the anaerobic threshold using the wearable air purification apparatus 100 during physical exercise. This method may determine the user's anaerobic threshold with higher accuracy than currently available non-laboratory approaches and, moreover, may allow said user to measure progress and improvement and, also, target an exercise intensity level that fits them.

The method includes a first step 3110 of generating a first dataset by detecting output from the pressure sensor over a time period during physical exercise. As set out above, the pressure sensor 152 is arranged to detect pressure changes caused by the user's breathing. The output from the pressure sensor 152 over the time period is recorded, for example in the onboard memory 154. The time period may be any time period during which physical exercise is carried out and, in some examples, may include the whole duration of continuous time period during which physical exercise is performed by the user.

The method includes a second step S120 of analysing the first dataset to monitor the user's breathing rate over the time period. In this example, the breathing rate is breaths per minute.

The method includes a third step 8130 of determining a breathing rate value indicative of the user's anaerobic threshold based on a change in the user's breathing rate over the time period. Without wishing to be bound by theory, it is believed that when passing the anaerobic threshold, lactate accumulates in the blood and causing the user's breathing to intensify. Therefore, by analysing the user's breathing rate over the time period and, in particular, the change in the user's breathing rate over the time period, it is believed that the breathing rate at which the user is estimated to reach the anaerobic threshold, i.e. the breathing rate value indicative of the user's anaerobic threshold, can be determined.

The described method of determining the anaerobic threshold using the wearable air purification apparatus 100 during physical exercise can be repeated at a later time in order to track the anaerobic threshold of the user. Since the anaerobic threshold is expected to change as the user's fitness changes, this may enable the user to track the change in fitness and to adjust exercise intensity if desirable. For example, a first breathing rate value indicative of the user's anaerobic threshold may be determined at a first time, and, optionally, a second breathing rate value indicative of the user's anaerobic threshold may be determined at a second time after the first time.

Hence, by utilising the apparatus 100 including the pressure sensor 152, the anaerobic threshold of the user can be determined, allowing for many potential use-cases such as digital coaching. This also allows access to another metric with which the user can track progress and which may not be available in known wearables given the limitations of known wearables in obtaining the breathing rate of a user when awake/during exercise as opposed to when asleep.

Figures 5, 6, 7, 8, 9 and 10 illustrate how the sensor data can be analysed and the breathing rate value indicative of the user's anaerobic threshold determined. Without wishing to be bound by theory, it is believed that breathing rate is interlinked to physical and perceived exertion such that breathing rate will encounter an elbow point as the body passes the anaerobic thresholds at a certain exercise intensity and, by taking the breathing rate during exercise, this threshold can be determined.

Figure 5 is a graph of the raw dataset output by the pressure sensor 152 over a first exercise time period.

As such, Figure 5 shows a change in air pressure as detected by the pressure sensor 152 over time, corresponding to successive inhalations and exhalations of the user. The raw output of the pressure sensor 152 may further includes other signals, such a high-frequency signal not indicative of the user's breathing at a comparatively low frequency.

Figure 6 is a graph showing the dataset plotted in Figure 5, wherein a low-pass filter has been applied to the raw dataset of Figure 5 to smooth the signal. Using the low-pass filter, the presence of the high-frequency signal has been reduced such that a sinusoidal signal emerges more clearly, and which may be more obviously indicative of a breathing signal.

Figure 7 is a graph corresponding to the graph of Figure 6, wherein a peak-finding algorithm has been utilised to identify peaks within the dataset, and which are indicated in Figure 7 by crosses. The peak-finding algorithm was configured to identify local maxima corresponding to time points where the detected pressure reaches a maximum. In this example, a total of seven peaks was identified by the peak-finding algorithm over the first exercise time period, indicative of seven user breaths over this time period.

By identifying the user breaths in the sensor data, the rate of change in the user's breathing over the exercise time period is identified. In this example, the breathing rate is computed as breaths per minute.

Figure 8 shows a graph illustrating the breathing rate (breaths per minute) for a second exercise time period. The second exercise time period is longer than the first exercise time period. The vertical axis (or y-axis') is the breathing rate in breaths per minute and the horizontal axis (or 'x-axis') is time, which can be considered equivalent to exercise intensity, since exercise intensity increases as exercise duration is increased.

Figure 9 shows a polynomial trend line, which is generated by fitting the polynomial trend line to the data of Figure 8.

Figure 10 shows the graph of Figure 9 but additionally includes a dashed line representing the user's breathing rate indicative of the user's anaerobic threshold.

The anaerobic threshold may be identified as an elbow point in the user's breathing rate over the second exercise time period. The elbow point is understood to be a local maximum of curvature of the plot. A person's breathing rate will encounter an elbow point as the body passes the anaerobic threshold at a certain exercise intensity. Hence, the breathing rate during exercise can be used to determine the anaerobic threshold.

In summary, the described method includes obtaining a pressure signal from the pressure sensor 152 of the head-worn wearable apparatus 100; smoothing the pressure signal using a low pass filter to better express the breathing signal recorded by the pressure sensor 152; using a model or algorithm to determine the breathing rate from the smoothed pressure signal; determining the breathing rate value indicative of the anaerobic threshold from a change in the breathing rate.

In this example, the method further includes using a trained machine-learning model to identify subsets of the dataset generated by the pressure sensor 152, where these subsets correspond to non-breathing events. The non-breathing events are events which result in pressure changes but are not considered indicative of the breathing or the breathing rate of the user. Non-breathing events may include, for example, sneezing or speaking.

In such examples, monitoring the user's breathing rate over the time period includes: determining whether a subset of the dataset is indicative of a non-breathing event using the trained machine-learned model which has been trained with labelled datasets; and excluding the subset indicative of the non-breathing event when determining the breathing rate value indicative of the user's anaerobic threshold.

Thus, additional functionality in this example includes a classification model to determine whether the pressure signal is a non-breathing event e.g. sneezing, coughing, talking, etc. Figure 11 illustrates a method of training the machine learning model. The method includes a first step S210 of generating multiple reference datasets. Each reference dataset is respectively generated by monitoring output over a time period from a pressure sensor 152 configured to detect pressure changes caused by a user's breathing.

The method of training the machine learning model further includes a second step 8220 of producing labelled training data by labelling portions of the reference datasets indicative of non-breathing events to associate those portions with non-breathing events.

The method of training the machine learning model further includes a third step S230 of using the labelled training data to train a machine learning model for identifying a non-breathing event in the dataset. Figure 12 illustrates a method of digitally coaching the user relative to the determined anaerobic threshold.

The method includes a first step 3310 of retrieving the determined breathing rate value indicative of the user's anaerobic threshold.

The method includes a second step 3320 of receiving output from the pressure sensor 152 to monitor a current breathing rate of the user.

The method includes a third step 8330 of notifying the user if the user's breathing rate approaches the breathing value indicative of the user's anaerobic threshold.

By monitoring the user's breathing rate relative to the breathing rate value indicative of the user's anaerobic threshold, it is possible to inform the user when approaching said threshold. Thus, the user may adjust exercise intensity, for example in order to ensure training below said threshold or training above said threshold.

Various changes may be made to the methods and apparatuses described herein.

For example, in the above-described examples, sensor data recorded by the onboard pressure sensor 152 is transmitted to the smart device 200 and subsequently to the server 300 for processing. In some alternative examples, some or all of the processing may be performed by the smart device 200. In some further examples, some or all of the processing may be performed by the air purification apparatus 100.

In some examples, the data transmitted to the smart device 200 and/or the server 300 may correspond to the dataset generated by pressure sensor 152 or may be data representative thereof. For example, the raw dataset may be transmitted, or the dataset following application of the low-pass filter, or the breathing rate over a period of time may be transmitted.

As set out in the above-described examples, a peak-finding algorithm is utilised for identification of the user's breaths over time in the dataset recorded by the pressure sensor 152. As an alternative, a negative peak finding algorithm may be utilised to identify minima, or any other means for identifying the user's breaths in the dataset The features disclosed in the foregoing description, or in the following claims, or in the accompanying drawings, expressed in their specific forms or in terms of a means for performing the disclosed function, or a method or process for obtaining the disclosed results, as appropriate, may, separately, or in any combination of such features, be utilised for realising the invention in diverse forms thereof.

While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.

For the avoidance of any doubt, any theoretical explanations provided herein are provided for the purposes of improving the understanding of a reader. The inventors do not wish to be bound by any of these theoretical explanations.

Any section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

Throughout this specification, including the claims which follow, unless the context requires otherwise, the word "comprise" and "include", and variations such as "comprises", "comprising", and "including" will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

It must be noted that, as used in the specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from "about" one particular value, and/or to "about" another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by the use of the antecedent "about," it will be understood that the particular value forms another embodiment. The term "about" in relation to a numerical value is optional and means for example +/-10%.

Claims

Claims: 1. A method of determining a user's aerobic-to-anaerobic threshold using a wearable air purification apparatus during physical exercise, wherein the method uses a pressure sensor of the wearable air purification apparatus configured to detect pressure changes caused by the user's breathing, the method including: generating a first dataset by detecting output from the pressure sensor over a first time period; analysing the first dataset to monitor a user's breathing rate over the first time period; determining a first breathing rate value indicative of the user's aerobic-to-anaerobic threshold based on a change in the user's breathing rate over the first time period.
2. The method according to claim 1, further comprising: notifying the user of the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold 3. The method according to any preceding claim, further comprising: detecting output from the pressure sensor to monitor a user's breathing rate over a second time period, which takes place later than the first time period; determining that the user's breathing rate approaches the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold; notifying the user if the user's breathing rate approaches the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.4. The method according to any preceding claim, further comprising: generating a third dataset by detecting output from the pressure sensor over a third time period, which takes place later than the first time period; analysing the third dataset to monitor a user's breathing rate over the third time period; determining a second breathing rate value indicative of the user's aerobic-to-anaerobic threshold based on a change in the user's breathing rate over the third time period.5. The method according to claim 4, further comprising: notifying the user of the second breathing rate value indicative of the user's aerobic-to-anaerobic threshold.7. The method according to any preceding claim, wherein analysing the first dataset to monitor the user's breathing rate over the first time period includes using a peak finding algorithm to determine the number of breaths over the first time period.8. The method according to any preceding claim, wherein analysing the first dataset to monitor user's breathing rate over the first time period includes: determining whether a subset of the first dataset is indicative of a non-breathing event using a trained machine-learned model that has been trained with labelled datasets; and excluding the subset indicative of a non-breathing event when determining the first breathing rate value.9. The method according to any preceding claim, wherein determining the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold comprises: transmitting the first dataset or data representative thereof to a remote server for the server to determine the first breathing rate value; and receiving from the remote server the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.10. The method according to any one of claims 1 to 8, wherein determining the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold comprises: transmitting the first dataset or data representative thereof to a smart device in communication with the wearable air purification apparatus for the smart device to determine the first breathing rate value.11. The method according to claim 10, further comprising: receiving from the smart device the first breathing rate value indicative of the user's aerobic-to-anaerobic threshold.12. A method of training a machine learning model for use in a method according to claim 8, wherein the method includes: generating multiple reference datasets, wherein each reference dataset is respectively generated by monitoring output over a time period from a pressure sensor configured to detect pressure changes caused by a user's breathing; producing labelled training data by labelling portions of the reference datasets indicative of non-breathing events to associate those portions with non-breathing events; using the labelled training data to train a machine learning model for identifying a non-breathing event in the first dataset.13. A method of digitally coaching a user relative to the user's aerobic-to-anaerobic threshold during physical exercise, the method including: retrieving a breathing rate value indicative of the user's aerobic-to-anaerobic threshold; receiving output from a pressure sensor of a wearable air purification apparatus to monitor a user's breathing rate; notifying the user if the user's breathing rate approaches the breathing value indicative of the user's aerobic-to-anaerobic threshold.14. A wearable air purification apparatus configured to carry out the method according to any one of claims 1 to 11, wherein the wearable air purification apparatus comprises: a fan assembly configured to provide a filtered airflow; a visor to direct the filtered airflow to a user who is wearing the air purification apparatus, a pressure sensor to detect the user's breathing.15. The wearable air purification apparatus according to claim 14, wherein the apparatus is configured to transmit the first dataset or data representative thereof to a remote server or a smart device.