GB2554417A - System for assessing effective physiological performance - Google Patents

Abstract

The method includes monitoring a user 2, for example respiratory rate or ECG, processing via data processing unit 6 to generate score 7 such as a heart-rate variance (HRV) score, and controlling display 8 of stimuli to the user, possibly including an animated graphic or information e.g. breathe more slowly, based upon the generated score 7. In embodiments, the invention monitors user HRV and provides an interactive biometric feedback loop to help the user control their diaphragmatic breathing, which may improve a user's control and reduce cognitive stress. Processing of the data may include time-domain or frequency-domain analysis on extracted subset(s) of the user data (e.g. heart beats or bpm) and/or comparison with a stored ideal reference, or combinations thereof. The mean squared differences of NN intervals (RMSSD) or FFT analysis of RR signal may be used. Envisioned users include emergency and military personnel or any user that may benefit.

Description

(54) Title of the Invention: System for assessing effective physiological performance Abstract Title: System for assessing physiological performance of a user (57) The method includes monitoring a user 2, for example respiratory rate or ECG, processing via data processing unit 6 to generate score 7 such as a heart-rate variance (HRV) score, and controlling display 8 of stimuli to the user, possibly including an animated graphic or information e.g. “breathe more slowly”, based upon the generated score 7. In embodiments, the invention monitors user HRV and provides an interactive biometric feedback loop to help the user control their diaphragmatic breathing, which may improve a user's control and reduce cognitive stress. Processing of the data may include time-domain or frequency-domain analysis on extracted subset(s) of the user data (e.g. heart beats or bpm) and/or comparison with a stored ideal reference, or combinations thereof. The mean squared differences of NN intervals (RMSSD) or FFT analysis of RR signal may be used. Envisioned users include emergency and military personnel or any user that may benefit.

At least one drawing originally filed was informal and the print reproduced here is taken from a later filed formal copy.

1/11

02 17

Figure 1

2/11

CM

Ο

CM

CM

o| o

col

3/11

02 17

4/11 o

CD

CD

Q_ cn

CD

CD i_

CD

CD

Q_

CD

CD

CL

1_

CD

Q.

Q.

_0 _Q

CD

CD .±=

CD

Q) .cn c CD CD

S— CD

CD _0 o

CD

CD i_ cn

CD

CD

CD

CL

CD

CD

CL &_

CD

Q_

Q_ cn

CD

Q_

CD

CD _0

CD

C

CD

In^-

CD

CD

Q_

O _Q

4—

O

CL

CL cn

CD

CD

Q_ _0

O _Q

CD

CD _O

C

CD i_

CD

Q_

Q_

CM

Ο

CM

CM > » CD 0

CD ff» CD § CL ' <

z-A 0 CD CD CD -0 CL 0 sCD O CD o

~ ο _o Ω-.Ω 0 .55 S 5 ⁵ * CD O

CD -0 CL 0 ζιλ ω

0) σ

CD _0

CD

CD

CD

TJ cn

CD

0)

CD _0

CD

CD

CD

Q_ cn o

Q_

Q_

O

CD

CD >

CD

CN

Q.

ω

0)

CD

CD

Q_

CD

CL

CN

Q_ ω

σ)

CD

CD

Q_

CL

CD o Q. S ^E

-Q 0 0 w £· _ CD CD > 0 CD O) b; CO £

CD a cl

E ω CD £

CD

CD i i_ x O ω c CD CD _Q CD >

CD _0

CD

CD

Q_ _O

C

CD i_

CD

Q_

Q_

CD c

O uri

CD _Q

CD c

CD £

CD _0 cn

CD

CD _Q

CN >» _Q

CD ·0 ~ ο 3 a.

E cn ξ 2 £ ω 2 £ = S ro ω ο £

S'o

O it o CO O -0

Nf *_ o 5 ί- 0

CD 0 c « — CD si.

CD Q0 CD Q.

0)

C '(f) o

CD

Q_

CD

0^-

CD

CD

cn
CD
0)
0
CD
_0
0
CD	- .
0	CD
CD >	0) 0 CD
CD	<r
0	o
4—	0
	o
CD
05	0
0	CD
CD	&_
_0	T0
0
0	CD
o	cn
[_	CD
o
CD	0
i_	0
T3	a
0	o
O	a
0	cn
o	0
	CD
	cn
CD	0
0) 0	CD &_
	>»
0	0
Q_	0
	CD
0)	0
0-
0	CD
-	1_
0	1_
0	CD
O
O	0)
o CD
o
	cn
0	CD
0	>»
CD	0
	CD
0	0)
0	0
_Q	T3
O	O >
0	CD
0
0	0
O
cn	cn
cd	in CD
	Q_
cn
CD cn	_o
o	CD
0	cn
0	CD
CD	j_7
cn 0 CD &_	0 it CD
0	.0
O >	cn
CD
O	o Z
CD	cn
O _CD	1-
Q_
.0	cd
CD	E
CD	cn
	0
cn	o
2
CD	O
cn	CD
CD	0)
4—
-	CD
4—
o	O
CD	O
Q_
0
a
o
<
co

(f)

CD

Q.

CD

Q_

E

CD cn

CD _Q

CD i_

CD 'cn

8e

2¾ o o 0) cn

CD

O o . cn .

o 8

CD CD i- e 0 8 8

8¾ cn cn cn

0 O Q. 0 CD

CD

0)

CD i_

CD >

CD

C

CD

O

CD

Q.

E _CD cn

CD >

CD cn

CD >

O i_

Q_

E o

CD _Q

0) _Q

O

CD i_

CD

Q_

CD

C

CD i_

O _0 ί 0 Ί c CD

O cn

0)

E

-- € > CD § °O 0 o ^c +- 0 g^D 8£ W cn £ 0 £ ω

5? (ΙΊ o

=3

CD

5/11

CM

Ο

CM

CM

a List.
o
		II
0
CD		X
0
0		CO
CD		CD
CD		0
i_ CD		CD >
ω CD		X
3
		CO
CD		CD
>		CD
0
0		CD	5
CD		CD > CD	O
E		0 0
CD		_0	>
if)	F	o	4—
>>			o
CO
CD		CD	_0
	ω CD
_0		0)
-·—1		0
O		CD	CD
0		0
—		CD
0		_0
Q_
0		o
		M—
are		0) _0
ω		CD
CD £		§
o		CN
0
CL
CL

ω CD to
o σ	E o
0	ω
s	>» ω
CD	CD
O	_0
CD	O
0	·—'
a	0
E	o
E	*0
o	Q_
o	0
o	—
CD	ω
o	CD
jz
0)	CD
0	>
o 0 CD	CL CX
CD	5
0)	CD
CD	0
—	CD
.ω	CD
tn	E
ω	-t—1
_cd cd	JZ O
0	CD
CD _0	CD
1—
	LL
CD	LL
>	0)
CD	0
0 CD	ω CO
_0	CD
§	O o
co	Q_

5 O 0 SZ >
0
0
CD
o’
CO
LO
1_
CD
_Q
E
0	O
0	CO
CD	LO
Q_ E	_0 0) 0
CD	o
ω
_0
co
	LO
_o	o
Q. LO
E	X
CD	CD
ω	0
.0
cd £	CD
o 0	ω CD
—	Q_
ex	E
ex	CD
	ω
ω
CD	o
	4=-
CD _0	1— LL
	LL
0)	CD
_0
	4—'
	O
ω σ	CD
c	0
CD	O
ω E	CD O
o	LO
to	CN
>»	ω
ω
CD	CD
_0	N
	'ω

co

Q.

O to

CD

CD

Q_

CD cr o

σ £

_o

Q_

E

CD ω

c

CD >

o

CD £

ω

CD

Q_

E

CD 'CD

CD ω

o

Ό

CD 2 ³

0· .5⁵

CD i_ 4—	ω CD
—	_0
CD	4—'
4— o	4— O
	0
to	CD
—	0
CD	O
ω	Q_
0	E
CD	o
>	o
'01
_	o
f	Q
co	o
Q. O	O 0 0)

CO — e£ o

CO ω

CD

CD

LO co

CD cd

Q

Figure 5

6/11

CM

Ο

CM

CM

Figure 6

7/11

02 17

	above
					co
				c		σ
				φ		C
φ				Έ		03
Ό				ο
Σ3		*		ο
		03
				>,		“-
Q.		C		ο		Ό
Ε		Φ		C		Φ
03 σ		Ε 03 Q		φ =3 CT		03 =3		ό
C 03 X		C =3 Μ—		ω Μ—		Ο 03 Ο		Φ E
φ		Φ		Ε		(/)		φ
σ		.C		—τ
C						C		.c
ω φ > φ		s above		ο ω δ·®		both poi		in 4 as t
		φ		c		w		Ό
φ		ο C		φ φ Ζ3 4=				Φ
σ		φ		Φ £>		φ		CO
C		=3			c		=3
03		cr φ		=3 φ ±±		φ		O 03 O
Ο	F	Μ—	F	£ Q-	F	c	F
C φ =3		ω Μ—		- Ε Ο 03		co _c		line
CT		Ο		c				ω _£= 4— r>
φ Μ— φ		φ		ο ω .5± φ Ο CD		g so
σ =3		ο Q.		Ω.Έ ™ φ		.c		Φ
1		Μ—			τϋ		Q.
Q. Ε 03		Ο φ CD 03		.c £ έ		4— c o w		O W Φ f—
ω		Φ		Έ		w Φ
φ		>						Φ
.C		03		Ο		CD		w
σ>		Φ		Ω_		Φ		Z)
1= φ _c		.C σ		φ Έ φ ο		ro Φ		in
		C			_c
σ		LL		φ
C					E
ί_|_		c\i		ΛΣΣ
				Φ		£
				03		φ
				Ξ3		CL
				Ο
				03		rt
			Ο
		CO

Figure Ί

8/11

Over Time

9/11

CM

Ο

CM

CM

,r~ ITJ ™ o

— m o ό

1~ (.r, o ©

1Λ O ©

Over Time onseA '4 n ay\f

10/11

Over Time11/11

02 17

dd

LO ό

SYSTEM FOR ASSESSING EFFECTIVE PHYSIOLOGICAL PERFORMANCE

Field of the Invention

The present invention relates to a system for assessing physiological performance and cognitive stress of a user. It is particularly applicable, but by no means limited, to monitoring a user’s heart rate variance and providing feedback to help a user control their diaphragmatic breathing, which can improve a user’s emotional control and reduce stress.

Background to the Invention

With the increase in numbers of mental health issues, especially in young people, systems and methods are needed that that can help in the prevention of mental health diseases caused by stress or anxiety. Diaphragmatic breathing techniques can be used as a means to control and reduce stress.

Diaphragmatic breathing is an established strategy to help manage anxiety and stress and is a core component of many techniques that are proven to be able to reduce stress and anxiety such as Cognitive Behavioural Therapy (CBT), mindfulness and yoga.

Heart Rate Variance (HRV) measures can be effective indicators of psychophysiological stressors and successful performance of diaphragmatic breathing.

The most effective methods of training and education are those which reward the user for correctly performing an action and further incentivises the user to continuously improve their performance. Envisioned usages of this invention can be for training and education of; young people, emergency services personnel, military, corporate professionals, and any user that may benefit from effectively improved emotional control, via such responsive emotional training simulation systems such as virtual reality environments or video games.

To achieve the best level of training and education, a relationship between heart rate variance and diaphragmatic breathing needs to be established, so that the development of metrics representing the relationship can then be used in a training system to provide feedback to a user.

The invention includes an interactive feedback loop between the system assessing a user’s

HRV and instructing/modifying the user’s diaphragmatic breathing, which complements and improves the user’s body's own biofeedback loop, thus allowing a user to better manage, control and improve their mental health and wellbeing.

Diaphragmatic breathing

Diaphragmatic breathing, also known as abdominal or belly breathing, requires the diaphragm to flatten during the inhalation cycle. This in turn causes the stomach to expand outwards whilst causing a partial vacuum. This partial vacuum allows air reach the lower sections of the lungs. The opposite is Shallow Breathing, or thoracic breathing, which draws a minimal amount of air into the lungs by using the intercostal muscles, instead of the diaphragm.

Most people unconsciously perform shallow breathing during the day. Shallow breathing is commonly associated as either the cause or the symptom of anxiety disorders, asthma and shock. Shallow breathing is also commonly associated to anxiety, stress, and panic attacks.

A full cycle of breathing consists of an inhalation then holding the breath, an exhalation and holding after exhalation.

Measurement of Heart Rate

Heart Rate Variance (HRV) is the phenomenon of the heart rate rising and falling on different time scales.

Heart Rate Variance (HRV) measurements, represent the oscillation in the interval between consecutive heartbeats, directly related with the oscillations between consecutive instantaneous heart rates. This is often referred to as cycle length variability, heart period variability, RR variability and RR interval tachogram (where R is a point corresponding to the peak of the QRS complex of the electrocardiogram (ECG) wave; and RR is the interval between successive R’s) (the QRS complex is a name for the combination of three of the graphical deflections i.e. the visual representation of sinus node depolarization, seen on a typical electrocardiogram).

HRV is any measurement that effectively measures the interval between consecutive beats and its variation. Other units used are RR (R-to-R) time, RMSSD (Root Mean Square of the Successive Differences), NN (normal to normal) time and instantaneous Heart Rate.

Heart Rate (HR) is the number of beats per minute (bpm) or the number of R activities in the QRS complex in a minute.

RR is time from R-to-R, in seconds e.g. the time between the R components (peaks) of two different QRS complexes (2 heartbeats).

RMSSD is a qualitative, statistical analysis of HRV values.

NN is the time between heartbeats, sometimes used instead of RR.

Instantaneous Heart Rate (IHR) is calculated from each R-R interval.

Baseline Heart Rate (BHR) is the average of all instantaneous heart rate over a period of 60 seconds.

The Responsive Metric (RM) used in the invention is the metric that represents the increase and decrease of HRV over time, it is not a medical standard. Instead, it represents a user's io performance and it is modified to provide a variable input representative of the quality of the users HRV.

Effects of breathing in humans

Good mental health can be defined as the ability to learn, to develop and maintain relationships with others and to cope with and manage change or uncertainty.

is Diaphragmatic breathing is a common feature of several different techniques of relaxation, from Yoga, to Tai Chi, to progressive muscular relaxation. This type of breathing ensures air reaches the lowest part of the lungs, and promotes a maximum exchange of oxygen for the carbon dioxide the body accumulates. This process, in return, lowers the heart rate and stabilizes blood pressure.

The autonomic nervous system (ANS) or involuntary motor system is regulated by the brain. The ANS, in turn, regulates the heart, digestive tract, lungs, bladder and blood vessels. The ANS is divided into two, opposing branches: The Sympathetic nervous system, which forces organs to react and do something (usually referred to as “Fight or Flight” system), and the Para-Sympathetic nervous systems, which causes the organs to enter a restful period (e.g.

Drop heart rate). The two systems introduce opposite effects of readiness and activity on the aforementioned organs.

In addition to the ANS, the RSA (Respiratory Sinus Arrhythmia) is a physiological phenomenon that reflects the acceleration and deceleration of HR when inspiring or expiring, respectively. While this effect varies from person to person, it has been observed that HR is closely related to RSA, and that HR increases as one's breathing becomes slower and deeper.

These are not the only systems or organs to impact on HR variation. It is proven that RSA exists even in the absence of breathing, and that other factors such as decrease in venous return (the amount of blood returning to the heart), Baroflex receptor reaction to changes in blood pressure, and many other organs and self-regulatory systems, as well as posture and head-up tilt (whether a person is standing up or laying down), will impact on a person’s HR variation.

When users carry out diaphragmatic breathing as instructed by a breathing coach, the resulting RR plot would exhibit an almost sinusoidal pattern in their HR variation (see Figure 11).

The measure of RR can be performed with ECG signals. RR is then calculated using the intervals between successive QRS complexes (QRS complexes result from sinus node depolarisation). From these readings, one can calculate the normal-to-normal (NN) intervals between heartbeats. These QRS complexes obtained during ECG readings give the RR interval (or the time between the R point of one QRS complex to the R point of the next, successive, QRS complex).

There are currently no large scale studies on HRV measures to accurately define what a normal value range is. In internally performed small user tests studies to capture and plot RR variation, to try and find a relationship between HRV and diaphragmatic breathing, it has become evident that diaphragmatic breathing has an effect on some of the Low Frequency (LF) components of RR.

Low Frequency content is not exclusively dependent on the breathing frequency or its depth, but its effects are so prominent that invention has four different metrics that rely on the amplitude and frequency content of an HRV measurement.

This means that even though it is clear that many other factors affecting the relationship and quality of the different frequency contents of an HRV measurement, the ratio between these for a person correctly performing diaphragmatic breathing whilst sitting down it can be safely used as a reference to develop a responsive metric.

Prior art

There are products such as Kubios and Bioforce, which offer software suites aimed at professional and amateur athletes. These apps support most standard HRV measurements to monitor and plot the athlete's performance and improvement throughout their training routines.

Kubios offers an array of metrics and statistics, both time-domain and frequency domain. RMSSD, NN50, Triangular index, Power Spectrum Density, Auto Regressive frequency domain analysis, and Welch periodogram for the Fast Fourier Transform (FFT) analysis and autoregressive models.

HeartMath have developed a system as described in (US6358201) based the variation of heart rate for the measurement of heart rate variability. Also described is how a qualitative metric (which they call the entrainment parameter or “EP”) can be derived from HR and that the metric could be used in software apps or games.

Metrics and methods

HRV can be measured using time-domain methods or frequency-domain methods. These can provide insight into a number of bodily functions, such as instantaneous heart rate secondary to respiration.

Physiological interpretations of RR intervals are not currently defined as disagreement exists in respect to Low Frequency content, and the lack of large scale user testing. Whilst there are some accepted relationships between RR and a series of bodily functions, not all elements present in RR have a meaning or a relationship.

Frequency domain measures for analysis of the RR interval tachogram are known and for the purpose of this document, Very Low Frequency (VLF) signals are smaller or equal than 0.04Hz, Low Frequency (LF) signals fit between 0.04Hz and 0.15Hz and High Frequency (HF) signals sit between 0.15Hz and 0.4Hz.

Time Domain metrics

Time domain metrics are less useful for this invention as they are usually employed for statistical time domain measures, collected over periods of time ranging from 5 minutes to 24 hours. The responsive metric needs to be instantaneous, giving the user feedback on their breathing as soon as possible.

However, RMSDD Time Domain metrics have been collected from user test data, to assist in discerning between good and bad HRV datasets whilst further developing the responsive metric. By analysing time domain metrics, it can be identified when a user has performed well or not so well, and for how long they sustained that level of performance. They can also help disambiguate readings that are more difficult to interpret.

As a standard, Time Domain HRV metrics are usually classified as statistical methods (e.g.:

RMSDD, or Root Mean Square of Successive Differences) or as geometric methods (e.g.

Sample Density Distribution) to display or represent HRV. These metrics usually display the amount of intervals above or below a specific time threshold (e.g.: 50ms for pNN50), or the density of certain RR values. These could be adequate metrics that display the overall performance for a user after a single session or multiple sessions.

Time domain metrics are useful as overall indicators of how a user has performed from one session to another, but are less useful as a real-time indicator.

Frequency Domain metrics

Frequency domain metrics are more useful for the present invention as they allow for instantaneous feedback.

Time constraints due to the nature of the envisioned sessions (which consists of 2-minutelong situations that generate very small data sets) and the need to extrapolate data quickly (for use of the responsive metric), thus the frequency domain metrics of the invention use is nonparametric Fast Fourier Transform methods as they do not require information ahead of time.

Both Parametric and Auto Regressive methods would require longer data sets (e.g. hours’ worth of data per user) to successfully generate metrics. Contemporary non-parametric power spectral density methods are sufficient to analyse short-term HRV signals without transient changes of heart period modulations.

By plotting the frequency content of the frequencies contained in the HRV measurement, peaks clearly emerge that closely relate to fundamental frequencies. Via direct observation of user testing it has been concluded that these fundamental frequencies are directly related with the person's breathing activities.

When plotting Heart Rate Variation in real time, it can be observed how the breathing affects

HRV. Whenever a person breathed in or out deeply, the curve plotted by the HRV measurement over time would smooth and begin to resemble a high amplitude, low frequency sine wave.

Plotting the FFT at each point of RR variation with a window of 32 samples, illustrates that the 30 frequency content would consist of a distinct peak (at the person's breathing frequency) and most of the other frequency content would be greatly reduced.

Regarding the reliability of the RR information, RR is by nature irregular and erratic heartbeats can create QRS complexes with no RR. There are also issues associated with hardware sensors, namely inherent delays in capturing and processing data and an irregular output rate of data.

Frequency content from Frequency Domain analysis

The aim of using frequency domain analysis on HRV measurements is to decompose the HRV measurement into individual contents and see how they relate in intensity (amplitude). A spectral analysis of the waveform can reveal the activity of both the sympathetic or parasympathetic branches of the autonomic nervous system.

The High Frequency band is widely accepted as a measure of parasympathetic or vagal activity, and peaks in this band correspond to heart rate variations in relation with the RSA and the respiratory cycle.

The lower frequency, however, can reflect both sympathetic and parasympathetic activity. The parasympathetic influences are particularly noticeable in this region whenever an individual takes deep breaths, and it can be seen as peaks in the frequency spectrum analysis down to 0.05Hz.

Problems and limitations of existing systems

Existing “emotionally responsive” systems that aim to train, prompt and reward control of emotional state are not engaging or effective. This represents a significant problem because the user groups that suffer from low attentional control (concentration) are also most vulnerable to common mental disorders, such as stress and anxiety that are associated with poor emotional regulation. It is also a significant problem when creating realistic training environments, such that they can be configured correctly and to be responsive to changes in a user’s performance.

This problem extends to user groups, such as children and young people where those who experience mental health problems are more likely to have poor educational achievement, greater risk of substance misuse, anti-social behaviour, offending and early pregnancy. Poor mental health in childhood and adolescence is also associated with poor health and social outcomes in adulthood. Another user group is emergency service personnel who require that the systems are dynamic such that their training reflects their real world situations this is also relevant to use in military organizations.

This problem can be found rooted in the conventional mechanics and interface devices employed within the existing products.

Designers and developers of emotional training systems have little or no access to technologies that would enable them to create dynamic, effective and realistic “emotionally responsive” training. Despite the existence of sensors to capture HR at a high enough resolution to be converted to HRV and therefore track levels of emotional control amongst users, there is nothing available to developers to harness this data through established mechanics or interface devices. This is stunting an area of development that can increase the enjoyment and benefits of emotional training and stress reduction.

A need therefore exists for a system that monitors emotional control that responds quickly and communicates effectively. Specifically, the system must sample the HR and produce useful data at a rapid speed i.e. providing useful data over the short periods that such training processes require and providing meaningful feedback about change in as little as 1 breath cycle, then based upon this data deciding on the most effective feedback to give.

is An example of the limitations of current systems using frequency based metrics is that due to the sample rate (1 sec) and the window length (the window length matches loosely to the breathing cycles). In order to increase the frequency resolution at such low frequency, it would be necessary to use large windows. However, increasing the window is not possible as this would prevent any real-time dynamic feedback, such as would be required in a real-time training simulation.

Using a 32 sample window which loosely relates to the time it takes for a user to take in 2 to 3 breaths would mean the metric would have a 30 second delay before giving accurate results.

Increasing the window length is not an option for two reasons:

It would result in a much longer delay than is allowable in the envisaged training in which the time is relatively short 2-5min will have only 200-300 samples, and;

The window would include several breathing cycles, losing the capability to demonstrate how the user is performing at that exact point and therefore not feedback dynamically.

As such, this severely limits the options to explore features of this metric. A finer resolution would be required to delimit areas for further analysis and be more forgiving. More effective techniques such as calculating how outlying the fundamental frequency from the optimal zone, or by limiting a region of the spectrum on which to look for local amplitude maxima would be required.

Therefore, one of the aims of the current invention is to provide a responsive metric system that can be fed into the training simulation mechanics, so that the environment can accurately responds to the user based on how the user is breathing over time.

This responsive metric will not represent HRV as per standard conventions. Instead, this metric represents the change in RR and HRV measures: it increases in value as HRV increases and the RR plot becomes more predictable, and decreases in value as HRV decreases, and the RR plot is less predictable.

The responsive metric should exaggerate the results in relation with the user's breathing performance in order to noticeably influence the training environment.

Summary of the Invention

Accordingly, the present invention provides a system for assessing physiological performance and cognitive stress of a user, the system comprising: a display configured to display a series of stimuli to a user; a user monitoring device, configured to receive an input from the user and provide an output, the output including; a stream of data for a period of time X; a processing unit including storage, configured; to receive the output from the user monitoring device, to extract one or more samples from the output of a duration Y; to create at least one subset of data from each of the one or more extracted samples, to analyse the subsets and to generate a score, to control the display of different stimuli from the series of stimuli by the display in response to the generated score. The advantage of analysing the analysing the multiple subsets and generate a score which controls the display of different stimuli in response to the generated score is a system which respond rapidly in real-time to provide meaningful information which can train a user response eventually helping them control their emotional state.

Within the present system the analysis of the subsets may comprise; time domain based analysis, frequency domain based analysis and comparing the processed multiple subsets with a stored ideal reference data set. It will be appreciated that the different analyses may be used in various combinations. The different methods of analysis allow for different stimuli to be chosen targeted at improving the user input.

The stored ideal reference data set may be obtained from previous user results. This will be appreciated as building a user profile during a session being transient or carried over between multiple sessions and stored permanently.

The user input to the monitoring device of the system may comprises one or more of; a heartbeat, an ECG voltage, a heart rate sample in beats per minute (BPM), a respiratory rate.

The output from the user monitoring device may comprises one or more of; a continuous electrocardiogram (ECG) record, a continuous plot of heartbeats or heart rates over time, a continuous plot of respiratory rates over time.

The output may be sampled in various techniques such as envelopes of data, as windows of data, as a data series, as a discrete approximation of a continuous signal.

The method of time domain based analysis may comprise; analysis of the normal-to-normal (NN) intervals in a continuous electrocardiogram (ECG) record, the analysis comprising the square root of the mean squared differences of successive NN intervals (RMSSD).

The method of frequency domain based analysis comprises; a frequency average based metric, which obtains the frequencies and respective amplitudes in a continuous electrocardiogram (ECG) that compose the RR signal via Fast Fourier Transform. The frequency domain based analysis may further comprise a linear regression based metric or a is fit to fundamental linear regression based metric. The frequency domain based analysis may be compared to stored target frequencies, which may be obtained from previous user results.

The generated score of the system may comprise, a Heart Rate Variance (HRV) score. The generated score may represent the increase and decrease of Heart Rate Variance (HRV) over time and can provide a variable input to the system to decide the displayed stimuli.

The display of the system may show an animated graphical representation of the user or avatar. The displayed stimuli may provide information to the user to help the user improve their generated score. The displayed stimuli may provide information to the user to adjust their diaphragmatic breathing techniques, such as “breathe more slowly” or “breathe more deeply” etc.

The user monitoring device of the system may have a buffer which receives an input from the user and provides output to the processing unit for a period of time Z before the first display of a stimuli to the user. The period of time before the display of a stimuli may be around 20 seconds. The processing unit can analyse the buffered output and generates a score to inform the choice of the first stimuli or provide the stored reference data and target frequencies. The generated score can be transformed by progressive iteration in response to the user’s performance creating a biometric feedback loop, this allows the user to train their emotional control.

The user monitoring device may comprise one or more of; a heart rate monitor, a greenlight sensor, stretch sensor, conductive fabric sensor, optical plethysmograph.

The processing unit may comprise at least one or more of; a computer, laptop, tablet, smartphone, video games console, web based application.

The present invention also relates to a method of assessing physiological performance and cognitive stress of a user, the method comprising; displaying a series of stimuli to a user; monitoring and receiving an input from a user; providing an output in the form of a stream of data for a period of time X; extracting one or more samples from the output of a duration Y; creating at least one subset of data from the each of the one or more extracted samples;

analyse the multiple subsets and generating a score; and controlling the display of different stimuli from the series of stimuli in response to the generated score.

The method may be a method of assessing physiological performance and cognitive stress of a user, the method comprising; displaying a series of stimuli to a user; monitoring and receiving a heartbeat input from a user; providing a continuous electrocardiogram (ECG) is record output for a period of time X; extracting one or more samples from the output of a duration Y; creating at least one subset of data from the each of the one or more extracted samples; analyse the multiple subsets according to one or more metrics and generating a heart rate variance score; and controlling the display of different stimuli from the series of stimuli in response to the generated score.

Brief Description of the Drawings

The present invention will now be more particularly described by way of example only with reference to the accompanying drawings, in which:

Figure 1 is a representation of the system of the current invention;

Figure 2 is a flow chart of the method of the current invention;

Figure 3 is a flow chart showing the analysis and score generation step of the method in more depth;

Figure 4 is a flow chart of an embodiment of the Time domain based analysis metric used in the invention;

Figure 5 is a flow chart of an embodiment of the Frequency Average based metric used in the invention;

Figure 6 is a is a flow chart of an embodiment of the Linear Regression based metric used in the invention;

Figure 7 is a is a flow chart of an embodiment of the Fit to fundamental linear regression based metric used in the invention;

Figure 8 is a graph showing how all the metrics behave on 32 window length;

Figure 9 is a graph showing how all the metrics behave on 256 window length;

Figure 10 is a graph showing how all the metrics behave on 1024 window length; and

Figure 11 is a plot of a dataset of RR values illustrating how the quality of the RR signal improves when performing diaphragmatic breathing.

io Detailed Description of Preferred Embodiments

Referring now to the drawings in general but particularly to Figure 1 the use and advantages of the invention will be described by way of example only.

According to one embodiment of the present invention, a system 1 for assessing physiological performance and cognitive stress of a user 2, the system includes a display 8 a user monitoring device 3 and a processing unit 6.

The display 8 is configured to output different stimuli, the stimuli presented are dependent on control from the processing unit 6, the stimuli can vary depending on the requirements of the system. It will be appreciated there are many kinds of stimuli that engage and provide feedback to a user 2.

The user monitoring device 3 is configured to receive an input 2a from the user 2 and provide an output 4 based on the input 2a to the processing unit 6. The user input 2a can be a physiological response related to the human body, for example, a heartbeat or respiratory rate. The user monitoring device 3 takes the physiological input 2a from a user 2 and outputs it in a form appropriate to be used by the processing unit 6, for example it will take the user heartbeat and create a continuous electrocardiogram (ECG) record. The data is output in a stream from the user monitoring device, the length of time of this stream is related to the training simulation that the system 1 is being used for. It will be appreciated that the stream of data could be interruptible or sent in bursts or packets.

At the processing unit 6 the stream of data output 4 is received, the data can then be placed in the available storage permanently or dealt with transiently. From the received stream of data, the processing unit extracts one or more samples. The duration of the samples can be variable, depending on the configuration of the processing unit 6 in relation to the analyses being used and the required resolution of the samples.

From the extracted sample the processing unit 6 can create one or more subsets 5 of data from each sample. The subsets 5a are analysed by the processing unit 6. The analyses of the subsets 5a can comprise; time domain based analysis, frequency domain based analysis, comparing the processed multiple subsets with a stored ideal reference data set or any combination thereof. The exact method of the analysis metrics is described in more detail with respect to their relevant Figures.

io From the analysis of the subsets of data 5a the processing unit 6 generates a score 7, this score 7 is the responsive metric, that represents a user's 2 performance (diaphragmatic breathing) and it is tuned to provide a variable input representative of the quality of the users HRV.

The processing unit 6 acting upon the generated score 7 controls the display 8 to display different stimuli. The stimuli presented will be relevant to the user 2 for improving their diaphragmatic breathing.

In an exemplary situation the system 1 is configured for an emotional training situation, a user 2 is connected to heart rate monitor 3. The system 1 starts with the display 8 of instructions to the user 2 to tell them to begin, such as to start breathing normally. The heart rate monitor 3 will be receiving the user’s input 2a, e.g. heart rate. The heart rate monitor 3 will output this as a continuous electrocardiogram (ECG) record 4. The processing unit 6 will receive this output 4 and begin to process the data. This will be done by sampling the stream as it is received, for each sample 5 the processing unit can create subset 5a to which can be applied one or a combination of the analysis metrics. From the analysis of the subsets a score 7 is produced, this can be a Heart Rate Variance score, this score 7 is used by the processing unit 6 to control what is displayed 8. For example, if the user’s breathing is shallowing and erratic caused by the emotional training situation then the display 8 may include instruction to concentrate on breathing slowly and deeply. The system 1 will continue to sample and analyse the output 4 from the heart rate monitor 3 throughout the duration of the emotional training situation, continually generating an updated score 7, which controls different stimuli displayed to the user 2 until the end of the training.

The system may have a stored ideal reference data set which can be preprogramed or can be obtained from previous user 2 results. The storage of the system can allow for a user 2 to build up a profile such that their results can be carried across various emotional training situations to continue their improvement of their emotional control. Such stored data could be interrogated outside of the system on a larger scale for further analysis.

The system 1 could operate using one or more inputs 2a from a user such as; a heartbeat, an

ECG voltage, a heart rate sample in beats per minute (BPM), a respiratory rate. In turn these will correspond with one more outputs 4 such as; a continuous electrocardiogram (ECG) record, a continuous plot of heartbeats or heart rates over time, a continuous plot of respiratory rates over time. The alternative inputs 2a and outputs 4 allow for various hardware to be used, while the system 1 will still perform the role of assessing physiological performance and io cognitive stress of a user 2 and providing effective timely responses via stimuli to improve the user emotional control.

The user monitoring device 3 of the system may comprise one or more of; a heart rate monitor, a greenlight sensor, stretch sensor, conductive fabric sensor, optical plethysmograph, it is envisaged that more than those listed could be used in the system.

The processing unit 6 of the system may comprise at least one or more of; a computer, laptop, tablet, smartphone, video games console, web based application, it is envisaged that the processing unit could take any suitable form to function.

The stimuli displayed are important in representing to the user 2 what they should do to improve their emotional control, e.g. by improving their diaphragmatic breathing. The stimuli may comprise an animated graphical representation of the user such as an avatar. Alternatively, stimuli may provide simply instructions to the user to adjust their diaphragmatic breathing techniques, such as “breathe more slowly” or “breathe more deeply” etc. It is envisaged that different emotional training situations will take advantage of a wide range of stimuli, such as, visual, audible, vibration etc. to communicate in the most effective way.

According to one aspect of the present invention, a method (as seen in Figures 2 and 3) of assessing physiological performance and cognitive stress of a user 2, includes the steps of; displaying a series of stimuli to a user 101, monitoring and receiving an input from a user 102, providing an output in the form of a stream of data for a period of time X 103, extracting one or more samples from the output of a duration Y 104, creating at least one subset of data from the each of the one or more extracted samples 105; analysing the multiple subsets and generate a score 106; and; controlling the display of different stimuli from the series of stimuli in response to the generated score 107.

The step of analysing the multiple subsets and generating a score comprises multiple sub steps (see Figure 3). The analyses of the multiple subsets may comprise one or more of, time domain based analysis 111, frequency domain based analysis 112, and, comparing the subset with a stored ideal reference data set 113.

The method can be repeated numerous times as shown by the return arrow 108 which takes the method from the step of controlling the display in response to the generated score 107 back to display the series of stimuli 101. The total number of times the method is iterated through will relate to the specific training situation that it is being used for example a time duration situation, a situation based on the number of breath cycles, or a situation requiring io maintenance of a good levelling of breathing control for a certain period of time.

In an exemplary use of the method of assessing physiological performance and cognitive stress of a user; comprises displaying a series of visual stimuli in respect of an emotional training situation for example, the display shows a visual representation of the user. As the user heart rate is monitored and output as an ECG record for analysis, the display will be provided instruction, e.g. continue breathing normal as illustrated by the visual representation. After the ECG record has been analysed and a Heart Rate Variance score has been produced, this is used to control the display, e.g. to update the visual representation and to encourage a user to change their diaphragmatic breathing. The method will then follow the loop to continue the emotional training situation, thereby the user gains continuous feedback in response via and in response to the stimuli to improve their diaphragmatic control and in turn their emotional control.

In Figure 2 a step of buffering a user input 109 can be seen, the buffering relates to a period of time where the monitoring device and processing unit are active but before stimuli are presented on the display. The time for the buffer can be in the region of 20 seconds, however, the period of time for buffer can be adjusted based on multiple factors. It is possible that the buffering period could be minutes or more to develop a more accurate user profile before beginning the method. The buffered output can be analysed using the same analysis as used on the multiple subsets to generate a starting score, before the emotional training situation begins.

The system and method allow a progressive iteration in response to the user’s performance thereby creating a biometric feedback loop, having the effect of allowing a user to better manage, control and improve their mental health and wellbeing.

Figure 4 shows the time domain based analysis metric used in the present invention. This metric consists of processing the data as it comes into the system, making a best effort to attribute a score value based on the lowest frequency present in the signal, excluding changes in the RR data caused by high-frequency transient signals, and augments low metric results in order to be more lenient with users that may be struggling with the diaphragmatic breathing.

The metric substantially follows the following operations:

1. RR intervals are input into the system and Values are added to a List.

2. Analysing previous Values in the List and determining if the RR intervals are increasing or decreasing.

io 3. Repeating Step 2 until the Values changes direction (i.e. wave peak reached). A Peak Value is stored as a variable “upperPeak” (if values are now decreasing), or into a variable “lowerPeak” (if values are now increasing).

4. Repeating Step 2 until the opposite peak is reached (e.g. if the Peak stored on step 3 was upperPeak, it waits until a lowerPeak was reached, and vice-versa).

5. Once an upper and lower peak have been stored, the amplitude of the signal can be calculated (difference between RR of both peaks) and the half the period (sum of the RR intervals for every sample between both peaks).

6. The Frequency is calculated from the period using F = 1/(peaklnterval*2*1000) (peaklnterval is multiplied by 2 because the difference between upper and lower peak represents only half of the waveform's period).

7. The Score is generated by calculating the log(freq) * amplitude.

8. A couple of “failsafes” are in place to avoid transient noises that should not be taken into account, e.g. putting a “hold” on direction change if the value change is too large or too small. This works, in effect, as a low-pass filter, avoiding any fast, higher frequency transients to count as a “direction change.”

9. The Score is then smoothed out:

9a. The Value is clamped to an average of the previous scores. The range of scores to be considered between the current sample and past scores is defined by the smoothing coefficient;

9b. If the score is not between an allowable margin of error above or below the average value, the score is replaced with the average of previous scores;

10. Finally, if the score is too low, it uses an anchor and a parachute value to bring the value back up. This improves the usefulness of the low metric (i.e. it makes the metric less “severe” for users that are not performing so well).

10a. Further normalisation is applied to ensure values sit between 0 and 1.

Figure 5 shows the Frequency Average based metric used in the present invention; all of the frequency domain metrics have as a starting point of obtaining the frequencies and respective amplitudes that compose the RR signal via Fast Fourier Transform (FFT).

io The metric substantially follows the following operations:

1. RR intervals are input into the system and Values are added to a List.

2. Waiting for the dataset to have at least X values (X = length of window).

3. Whenever the dataset is large enough to accommodate a window, it starts processing FFT each time a new RR value is inputted to the system.

is 3a. e.g. for data point 10, and a window of length 10, calculate FFT for points 1 through 10;

4. Repeat step 3 until the system ends (e.g. if the last RR interval sample is sample number 530, and window size is 25, calculate the FFT for samples at index 505 through 530).

5. The result from step 3 will give us a list of all frequencies and their amplitudes for the given sample window. Data is then set to ignore DC7 component of the signal.

Figure 6 shows the Linear Regression based metric used in the present invention; when a user approaches an ideal diaphragmatic breathing technique, exhibiting deep breathing and a state of relaxation (i.e. Synchronising their biofeedback systems), the RR signal resembles a sinusoid. The signal has little content other than the fundamental and some harmonic content. By finding out the difference in power (ratio) between the fundamental and the signal we can obtain a metric that, for most cases, resembles the regressive metric approach.

The metric substantially follows the following operations:

1. Finding the lowest possible frequency with highest possible amplitude, a frequency Z;

2. Calculating the difference between that frequency's amplitude (Z's amplitude) and the average of the signal power (minus the frequency Z);

3. The resultant ratio is the metric.

4. If there is not a prominent, high amplitude, low end frequency, a frequency will still be 5 chosen, but the resulting ratio is so low that the resulting metric value for that will be very low.

Figure 7 is shows the Fit to fundamental linear regression based metric used in the present invention. This technique uses the results of the FFT (as seen in Figure 5), and then analyses the FFT at any given point. The metric then performs linear regression fitting to draw the line that best fits every point in the FFT. This is based on the principle that with deeper, slower io breathing, the RR signal will demonstrate a large peak at about 0.05Hz-0.1 Hz, and almost no energy above that. The relationship between the highest amplitude frequency (usually related with RSA activities) and the rest of the signal becomes apparent.

The main advantage of this metric is that other values and data can be collected and used to tweak the metric, like evaluating goodness of fit, plot residuals or generate predictions

The metric substantially follows the following operations:

1. Find the highest amplitude frequency and retrieve its index and amplitude.

2. Find the average of power of all frequencies above the fundamental.

3. Calculate the centre point for this portion of the frequency spectrum (frequency content above the highest amplitude frequency.

4. Perform linear regression fitting so that the line fits both points calculated in 1 and 3.

5. Use the slope of the line calculated in 4 as the metric.

Figures 8, 9 and 10 are graphs showing how all the metrics behave on 32, 256 and 1024 window length respectively. It will be appreciated that the frequency response is quite blocky, due to the lack of frequency bins in the region of the spectrum being analysed i.e. there is low resolution. When increasing the window size, resolution is gained (the frequency response becomes less blocky) but the time domain accuracy dies down.

Figure 11 is a plot of a dataset of RR values illustrating how the quality of the RR signal improves when performing diaphragmatic breathing, as highlighted by the area C having an almost sinusoidal quality when, compared to the more erratic signals in the areas delimited by areas A and B.

There are multiple advantages associated with the system and method of for assessing physiological performance and cognitive stress of a user as shown in the above description and drawings, including but not limited to, providing significantly stronger motivations for users to learn and practice diaphragmatic breathing techniques as a form of emotional control compared to analogue and non-responsive metric solutions, and significantly reducing the barriers for designers and developers to create emotionally responsive training that is more compelling and emotionally intense as well as provide benefits for users wellbeing.

Claims (30)

Claims:

1. A system (1) for assessing physiological performance and cognitive stress of a user (2), the system comprising:

a display (8) configured to display a series of stimuli to a user (2);

a user monitoring device (3), configured to receive an input from the user (2) and provide an output (4), the output including;

a stream of data for a period of time X;

a processing unit (6) including storage, configured;

to receive the output from the user monitoring device (3);

to extract one or more samples from the output (4) of a duration Y;

to create at least one subset (5) of data from each of the one or more extracted samples;

to analyse the at least one subset (5) and to generate a score (7); and to control the display of different stimuli from the series of stimuli by the display (8) in response to the generated score (7).

2. A system as claimed in claim 1, wherein the analysis of the subsets (5) comprises; time domain based analysis.

3. A system as claimed in claim 1, wherein the analysis of subsets (5) comprises; frequency domain based analysis.

4. A system as claimed in claim 1, wherein the analysis of subsets (5) comprises; comparing the processed multiple subsets with a stored ideal reference data set.

5. A system as claimed in claim 1, wherein the analysis of subsets (5) comprises; a combination of at least two of; time domain based analysis, frequency domain based analysis and comparing the processed multiple subsets with a stored ideal reference data set.

6. A system as claimed in claim 4, wherein the stored ideal reference data set is obtained from previous user results.

7. A system as claimed in claim 4 or claim 5, wherein the stored ideal reference data set is transient.

8. A system as claimed in any previous claim, wherein the input from a user (2) comprises one or more of; a heartbeat, an ECG voltage, a heart rate sample in beats per minute (BPM), a respiratory rate.

9. A system as claimed in any previous claim, wherein the output (4) comprises one or more of; a continuous electrocardiogram (ECG) record, a continuous plot of heartbeats or heart rates over time, a continuous plot of respiratory rates over time.

10. A system as claimed in any previous claim, wherein the one or more samples from the output (4) comprise; an envelope of data, a window of data, a data series, a discrete approximation of a continuous signal.

11. A system as claimed in claim 2, wherein the time domain based analysis comprises; analysis of the normal-to-normal (NN) intervals in a continuous electrocardiogram (ECG) record, the analysis comprising the square root of the mean squared differences of successive NN intervals (RMSSD).

12. A system as claimed in claim 3, wherein the frequency domain based analysis comprises; a frequency average based metric, which obtains the frequencies and respective amplitudes in a continuous electrocardiogram (ECG) that compose the RR signal via Fast Fourier Transform.

13. A system as claimed in claim 12, wherein the frequency domain based analysis comprises; a linear regression based metric.

14. A system as claimed in claim 12, wherein the frequency domain based analysis comprises; a fit tot fundamental linear regression based metric.

15. A system as claimed in claims 12 to 14, wherein the frequency domain based analysis is compared to stored target frequencies.

16. A system as claimed in claim 15, wherein the stored target frequencies are obtained from previous user results.

17. A system as claimed in any previous claim, wherein the generated score (7) comprises, a Heart Rate Variance (HRV) score.

18. A system as claimed in any previous claim, wherein the generated score (7) represents the increase and decrease of Heart Rate Variance (HRV) overtime and provides a variable input to the system to decide the displayed stimuli.

19. A system as claimed in claim 1, wherein the storage is configured to store multiple instances of results obtained over time from a user (2).

20. A system as claimed in claim 10, wherein the display (8) shows an animated graphical representation of the user (2).

21. A system as claimed in any previous claim, wherein the displayed stimuli provide information to the user to improve the generated score (7).

22. A system as claimed in any previous claim, wherein the displayed stimuli provide information to the user to adjust their diaphragmatic breathing techniques, such as “breathe more slowly” or “breathe more deeply” etc.

23. A system as claimed in claim 1, wherein the user monitoring device (3) has a buffer which receives an input from the user (2) and provides output (4) to the processing unit for a period of time Z before the first display of a stimuli to the user (2).

24. A system as claimed in claim 23, wherein the period of time before the display of a stimuli is substantially 20 seconds.

25. A system as claimed in claim 23 or claim 24, wherein the processing unit (6) analyses the buffered output and generates a score (7).

26. A system as claimed in previous claim, wherein the generated score (7) is transformed by progressive iteration in response to the user’s (2) performance creating a biometric feedback loop.

27. A system as claimed in claim 1, wherein the user monitoring device (3) comprises one or more of; a heart rate monitor, a greenlight sensor, stretch sensor, conductive fabric sensor, optical plethysmograph.

28. A system as claimed in claim 1, wherein the processing unit (6) comprises at least one or more of; a computer, laptop, tablet, smartphone, video games console, web based application.

29. A method of assessing physiological performance and cognitive stress of a user, the method comprising:

displaying a series of stimuli to a user;

monitoring and receiving an input from a user;

providing an output in the form of a stream of data for a period of time X;

extracting one or more samples from the output of a duration Y;

creating at least one subset of data from the each of the one or more extracted samples;

analyse the multiple subsets and generating a score; and controlling the display of different stimuli from the series of stimuli in response to the generated score.

30. A method of assessing physiological performance and cognitive stress of a user, the method comprising:

displaying a series of stimuli to a user;

monitoring and receiving a heartbeat input from a user;

providing a continuous electrocardiogram (ECG) record output for a period of time X;

extracting one or more samples from the output of a duration Y;

5 creating at least one subset of data from the each of the one or more extracted samples;

analyse the multiple subsets according to one or more metrics and generating a heart rate variance score; and

10 controlling the display of different stimuli from the series of stimuli in response to the generated score.

Intellectual

Property

Office

Application No: GB 1616326.3