Friday, August 31, 2018

Fast Start Strategy revisited and expanded

One of the most promising modalities to improve performance of a short time trial or equivilant is the fast start strategy (FS).  In a previous post this was reviewed but precise heart rate and respiratory data were missing on my end (no Hexoskin).  I thought it may be interesting to repeat some of the various permutations with accurate HR, ventilation as well as costal/leg SmO2 measurements.  The excellent study above discusses potential mechanisms of why this protocol may cause performance gains.  I will try to look at similar themes as well as some additional observations.  Finally, the "cost/benefit" of employing this type of power pattern is briefly reviewed.

To start with, I would like to briefly explain a few physiologic parameters of interest.

This is simply the amount of oxygen the tissue is utilizing.  VO2 max would be the maximum O2 usage attainable.
The classic definition is derived from the Fick equation:
VO2 = Q × (CaO2 − CvO2)
Q is the cardiac output which is (heart rate x stroke volume).  The stroke volume is the amount of blood pumped per beat.  As exercise intensity rises, the stroke volume rises, but plateaus well below the max heart rate.  At high work rates, we can assume stroke volume as a relative constant (technically it could even drop at very high HR from inadequate atrial filling).  That would mean cardiac output (at high work loads) is very proportional to heart rate.

(CaO2 − CvO2)
This refers to the difference in arterial and venous O2 content.  Although not a direct measurement of such, the NIRS O2 saturation is related to this somewhat, discussed next.

Locomotor muscle O2 saturation
Both oxygen delivery and consumption are able to affect the measurement of SmO2.  The balance between delivery increase (vasodilitation) and higher consumption (motor activity) may yield a variety of final results.  The net balance between the delivery and consumption is reflected as the amount of extraction.  For example if the delivery stays constant, a higher consumption (exercise) will cause a larger O2 desaturation (and a higher net extraction).  On the other hand, a decrease in delivery (heart failure) with stable consumption can lead to higher extraction and lower measured tissue O2 as well.  Both conditions lead to higher extraction though different mechanisms.

It has been shown that at moderate exercise intensities both HR and ventilation are related to the VO2.  Certainly at work rates above the MLSS, ventilation will be boosted from acidosis compensation.   This is a figure from a study looking at this relationship.

The end result is that looking at the early ventilation rise in either Fast or Slow starts may be a reflection of VO2 on kinetics.

Does tHb indicate blood flow?
Although it is tempting to use this as a measure of flow, it just does not work out.
There are some sophisticated methods to estimate flow (without cuff occlusion), but the simple tHb tracing can be misleading. 
From the quoted paper:
Handgrip exercise:

With exercise the flow is up (L panel), SmO2 is down (R panel) but tHb is flat.  This would be the general situation in most of our tracings, except for the consequence of external compressive forces.

With Cuff occlusion:
The tHb is relatively flat in cuff occlusion, but does show a tiny rise with cuff release and compensatory hyperemia.  The cuff is not surrounding the area of measurement so no compressive effects are seen.

Given the above background, several physiologic differences will be looked at in the Fast vs Slow start routines.  
  • Is there better O2 extraction in locomotor muscles?  
  • Is there a difference in cardiac output as reflected by heart rates (assuming constant stroke volume)?
  • Is there a difference in early ventilation increase (as a potential indirect look at VO2)
  • Is the end interval ventilation rate different?
  • Is the post interval ventilation recovery faster in FS?
  • Re examination of the costal O2 desaturation pattern in view of now having ventilation data.
  • Can a Fast start prime the "system" resulting in faster recovery than otherwise seen in an Even pace

VL desaturation patterns 
Slow vs Fast start 3 min at 320 watt average total.


  • Both intervals started at the same O2 saturation (65-66%) and since this was done on the same ride, assume the position did not change.
  • The end O2 sat was about the same (53-54%)
  • There was a faster and deeper drop in saturation in the Fast start.  At 20 seconds into the effort, the Fast start was down a net 5% over the Slow start effort (62 vs 57%).  The maximal desaturation was lower in FS (51 vs 53%).
  • A slightly faster re oxygenation to a similar peak post interval in FS (55 vs 50 sec).
When I initially saw this my first thought was that although the O2 extraction was theoretically higher in the Fast start, perhaps it was simply related to higher levels of muscle contraction force (from the high power of the FS) causing external compression (and reduced flow/supply).
For instance the following is a total Hb tracing of the latissimus muscle during a heavy load pulldown.  The prompt reduction in tHb (blue down arrow) is the start of muscle contraction and the blue arrow up is the end of the exercise set.

Does this occur during the Fast or Slow starts?
Here is the Slow start VL tHb (in purple):

 And the Fast start VL tHb (in purple):
Nearly identical!
There does not appear to be a compressive blood volume reduction.
Therefore it does appear that the O2 extraction is faster and more complete in the initial part of the Fast start interval unrelated to external compressive flow restriction.

Does the RF have a similar pattern?
Here is the Slow start RF O2 tracing:

 The Fast start:

  • The O2 saturation drop is faster and deeper in the Fast start than the slow start (50 vs 42 % at 20 sec).  Similar to the VL.
  • The end interval saturation is similar in both

Is the tHb pattern different?

This is a graphic of the zoomed out ride (power in gray).  Both tHb patterns look similar (red color).
Zoomed views
Slow start:
 Fast start:
 The RF patterns are similar in both intervals, as in the VL monitor tracings.

Summary observations:
  • Both the RF and VL desaturate faster as well as reaching lower total values in the Fast start interval
  • There is no notable difference in tHb measurements between Fast and Slow starts
  • Although this is not meant as a flow measurement, it does seem that there is at least a more rapid O2 extraction (with higher peak extraction) with a more significant hypoxic stimulus for metabolic change in the Fast start.  One of the hypothesized mechanisms for FS superiority is a more rapid buildup (or loss) of biochemical factors associated with muscle hypoxia.

Respiratory parameters:
The Hexoskin shirt has been shown to give accurate measurements of breathing rate, index of minute ventilation (when not calibrated to the individual) and heart rate.
Is there a difference in any of these parameters?

Slow start:
Fast start:
There appears to be several differences between intervals
  • The heart rate rose faster in the Fast start.  The 30 sec HR was higher in the Fast start (149 vs 141) supporting the concept of a faster rise in cardiac output using that technique.
  • The peak heart rate (at the end) was higher in the slow start (170 vs 166).
  • Minute ventilation also rose faster in the Fast start (98 vs 86 at 30 sec), but the end of interval value was similar (208).  If ventilation is a valid marker of VO2, the higher rise in the Fast start supports the concept of faster VO2 on kinetics
  • Post interval ventilation (one minute after 170 vs 118 while coasting) is markedly higher in SS.  Since ventilation more quickly returns toward "normal" in FS, readiness for another heavy load occurs sooner.  This may be related to the higher overall oxidative metabolism in the FS with less O2 debt.

What about average heart rate? 
Unfortunately, this was not easily obtainable given the problems in parsing the Hexoskin data (I did not record it as usual on Ipbike - see last post) but it was approximately 4 BPM higher in the Fast start.

For a better look, here are two intervals (on different days) with solid average HR data:

  • The average heart rate on the Fast start interval was 6 BPM higher.
  • The pattern of faster HR rise was also seen again
A look at costal O2 dynamics.  
The costal O2 potentially reflects both cardiac output redistribution effects, as well as increased respiratory muscle work rates (higher O2 usage by those muscles) from ventilation rise (normal O2 needs as well as lactate buffering).  This paper was reviewed awhile back but is an important theme for the data below.

Total Hb
Slow start costal tHb:

Fast start tHb:
  • The tHb patterns of both Fast and Slow starts are very similar at the costal muscle location.

Costal O2 desaturation tracings:
Slow start costal O2

 Fast start costal O2
  • Although the initial desaturation rate is faster in the FS, the ultimate final value is lower in the SS (25 vs 30%).
  • More importantly, the O2 sat curve is relatively flat/stable in the FS, but steadily dropping into historic minimal values in the SSThis occurs despite the similar ventilation rates at the end, (seen with the Hexoskin above).  The end interval costal hypoxic pattern in the SS is associated with shortened time to fatigue (seen in other intervals like a 1 min max).
  • The similar ventilation rates (at the end) should equate to similar muscular work loads of the respiratory muscles.  Therefore, the lower costal O2 values end interval may be more reflective of redistribution effects.  Since power to the legs is higher at end interval SS, the redistribution of cardiac output may be further diverted away from the costal muscles.  It could also be possible that the "efficiency" of these muscles become impaired with hypoxia (in SS) leading to reduced respiratory function and respiratory fatigue (leading to severe overall decompensation with continued effort).
  • Recovery time (black arrows) is shortened in FS.  Return to a more stable baseline takes longer with the SS.  This may be related to the higher ventilation rate seen post interval in the SS (higher costal muscle O2 consumption/extraction).  
  • The above potentially could hamper re oxygenation kinetics of the locomotor muscles as well (metobo-reflex - sympathetic vasoconstriction of the leg vasculature).

A look at longer intervals:

Given the observation that both HR, ventilation volume and presumably VO2 on kinetics are improved with the initial FS pace, does starting a longer interval this way adversely impact the remainder.  For instance, in a hypothetical race, are we worse off with starting with a 30 second effort at a higher power before dialing back to about MLSS vs just keeping at the MLSS from the start?  Does that initial 30 sec power burst "cost" us later, or does it create the optimal exercise physiology environment for the upcoming interval?

The best Hexoskin equipped intervals that I have to compare are as follows (not perfectly matched):
This is a 7 min Even start at 251 watts:
And the respiratory data:

Compare it to a 10 min 253 watt interval after an initial 30 sec at 400 watts:

  • Baseline HR about the same 
  • HR at 30 sec higher in FS (148 vs 134)
  • Steady state HR about the same
  • Baseline ventilation was the same
  • Ventilation at 30 sec higher in FS (95 vs 72)
  • Steady state ventilation toward the end about the same in FS vs ES
  • One min post interval ventilation lower in FS despite an extra 3 minutes and the higher initial power (however the coast time was shorter in ES)
  • Costal O2 pattern similar (given that the baseline values and interval length were different)

Some final thoughts (open to corrections)

Is there better locomotor muscle O2 extraction in either condition?   
  • Yes, it's higher in FS
Is there a difference in cardiac output as reflected by heart rates (assuming constant stroke volume)?
  • Yes - It appears that both average and 30 second heart rates are higher in FS.  Final HR generally is not.
Is there a difference in early ventilation increase (as a potential indirect look at VO2)
  • Yes, it's higher in FS
Is the end interval ventilation rate different?
  • Both conditions are very close
Is the post interval ventilation recovery faster in FS?
  • Yes
Re examination of the costal O2 desaturation pattern in view of now having ventilation data.
  • Costal O2 at the end of interval is higher in FS despite similar ventilation rates.  This raises the question of cardiac output redistribution(with lower costal flow and higher extraction) and resultant changes in respiratory muscle efficiency/fatigue.
Can a Fast start segment prime the "system" resulting in faster recovery than otherwise seen with an Even pace?
  • Possibly (quicker ventilation recovery)
How much does a Fast start cost us (fatigue, end ventilation/HR elevation)?
  • Not very much - end HR, end ventilation, end costal O2 are all equivilant

As an end comment, these observations were done on one person, older than the usual test subject, and genetically more of a strength than endurance athlete.  Whether or not younger, more elite endurance folks would show this pattern is unknown.  However, with relatively simple gear, these same test intervals could be done by anyone to see if they would benefit from a Fast start strategy.

 VO2 max related posts

Wednesday, August 29, 2018

Hexoskin - review on the road

 Hexoskin - review on the road

One of this blog's main focus is the monitoring of physiologic data in the field while exercising.  To expand on that, I recently used the Hexoskin shirt on a road ride with the supplied software on an android phone.  As discussed previously, the unit provides consistent readings of heart rate, ventilation parameters and activity.  After the ride one can download the results and look at them in the comfort of your home (what I had been doing). 
However, as seen in some prior posts, the ventilation data correlates well with the MLSS, RCP and beyond.  From a practical standpoint, can we use this device to help with pacing, racing, testing in real time.  In addition, some sports (XC skiing), do not have power meters.  It would be potentially quite helpful to know your minute ventilation (after MLSS/ventilation calibration on a running test for instance) on the course for optimal pacing.
Off the bat, one downside of the Hexoskin is that there is no Garmin datafield, so no watch/cycle computer support.
All we have is the smartphone app which follows.

This is what the android app screen looks like:
This is blown up of course.

  • The screen will dim and turn off (no screen lock in your app) unless the time out is altered (and still not a real) fix.  The best I could do was a screen time of 30 minutes then it went to sleep.  Yes, it's a PITA to have to put a password in to get it back on.
  • Wasted screen real estate (see the top 1/3).
  • No full screen option (note the nav bars, notification bar)
  • For real time monitoring the font size is too small to easily read and the important metric of ventilation is not emphasized
  • The distance/time is not even needed since the user will have a bike computer
  • It would be nice to have a running graph on the bottom
  • No full screen option
  • No high brightness mode
  • No pause, laps
  • Difficult to press the tiny buttons with gloves, or while moving 
  • What happens in the rain - no screen lock
Bottom line - barely usable 

As an example of proper UI for functionality is Ipbike for Android:

  • User defined fields (data, averages, min, max, time spans)
  • User defined graph
  • User defined font size
  • Multiple screens possible (swipe across).  So different screens for different needs (whole ride vs intervals vs a max value screen)
  • Laps, pause buttons
  • Big buttons
  • Screen lock (for when it's raining)
  • High brightness and screen-on lock (no sleep)

Next gripe, Hexoskin data analysis:
If one uses the Hexoskin phone app, the bluetooth protocol is such that either the same or different phone will be unable to read the heart rate in another app.  So your Garmin watch, android phone (ipbike) won't pick up the heart rate while the Hexoskin app is communicating with the shirt.  That's ok and I realized the situation pre ride, but that meant that I did not have HR data in my ride file, it was in the Hexoskin file.
The problem with that is although you can look at the data on their website, their is absolutely no analytical capability there.  So no average HR during an interval which was what I needed.
However, the "help FAQ" mentions that you can export a .csv and work off that data.
Let's look at that .csv:
It's a mess!
There is valid heart rate data in there but it's also full of "crazy" values and zeros.  We could write a script to ignore the zeros but not the other values since they are occasionally close to the real ones.
In addition the ventilation is in different units (?corrected for weight and size) and multiplied by 1000 (?)
I am sure there is method in this madness, but "I'm a doctor, not a mathematician".  In addition, the end user just should not be subjected to this degree of user unfriendliness.

I did find a site called FitnessSyncer that will import the Hexoskin file directly from the parent web data (no download needed).
Here is the graph:

The problem is that you can't zoom, nor get averages (at least in the free version)

Next workaround:
Export the FitnessSyncer file as a .fit, then import that into my trusty site:
It seems that on a more granular level the "crazy values" are still in there.  In addition to the extra values, the time was off with the ride about 1 hour longer than it should have been.

There is also a program recommended by Hexoskin called Vivosense that can read the raw data - but it costs over 2000 dollars!  I downloaded a trial version and still failed at getting an interval average HR.  

Bottom line:
  • The Hexoskin shirt is a product with great promise but extremely poor implementation.  From my perspective, the potential usefulness of the data for endurance exercise is tremendous.  Unfortunately, the data interpretation is hampered as discussed above.
  • If you use the app, the ability to analyze the heart rate data is lost (except in view form). 
  • The app itself was evidently written by folks who meant well but never used it on a outside bike.
  • The Hexoskin web data display has very limited (really none) analysis ability.
  • No Garmin data fields for ventilation parameters, limiting it's use in running, skiing in real time.

Monday, August 20, 2018

MLSS +10 watts, a journal review plus personal data

A couple of posts ago the effects of cycling at MLSS plus 25 watts (10%) was discussed.  Recently a study was published looking at the effects of cycling at MLSS plus10 watts (3-5%).  I am going to go over some of the details and findings.  I would also like to express my appreciation to the team of researchers who did this valuable study.

One of the reasons of doing this type of study is that the steady state maximal power definition is just that - steady state.   It is well recognized that power just above this leads to decompensation, acidosis and shortened time to fatigue.  In practical terms, avoiding power in this region for extended time could be a priority in a given race.  

One of the conclusions of the study is the fact that once the MLSS+10 trial was done, the subjects were "spent" and had limited reserves for available power/time.  So not only is it important to know your MLSS, but it would be quite helpful to monitor some sort of response in real time to prevent the beginnings of decompensation.  In terms of getting an accurate MLSS, the study team did multiple 30 min time trails with blood lactate measurements.  Even so they estimated that the error could have been near 9 watts.  Given the precision needed, encouragement is given to readers to use the MLSS method in a previous post that combines limited testing, no 30 min intervals, but excellent accuracy.  From the monitoring standpoint, the use of NIRS, resp rate/ventilation, and heart rate overlaps what has been discussed here.  Surface EMG and metabolic cart parameters were also done.
The basic procedure was for the subject to cycle at 30 minutes at either the MLSS, MLSS+10w and then do a time to exhaustion trial (after a 3 min rest then riding at 80% peak PO).

The Data:

Muscle O2, measured as deoxyhemoglobin:

Interesting observations:
  • The pattern of desaturation is different in the RF and VL.  The VL reaches a maximum desaturation within 5 minutes and remains levels whereas the RF continues to drop throughout.
  • The VL desat is therefore not helpful in demarcating these zones.
  • The RF desat is theoretically helpful but on a practical basis (with measurement changes from sensor position, temperature, hydration) is probably not useful in real time.  In other words if I was on my bike I could not easily use this to stay out of MLSS+10.  The 3 point change may be lost in the noise.

Heart rate:

A slight but significant change in HR.
But can it be reliably used in the field given drift, temp, hydration?

Respiratory rate and ventilation:

Both ventilation and respiratory rate show a large differential between MLSS and +10w.
Although not something routinely used, a very intriguing potential modality to employ in real time.  Also reminiscent of the MLSS and MLSS +25 watt comparison I personally did.

Here is an interesting look at the end trial, time to exhaustion stats:
The lactate, HR, RPE were the same in each end of TT "sprint".
However, both the RR and ventilation were down in the MLSS+10.  According to the authors of the study, the drop in peak ventilation may have simply been due to the decrease in exercise time preventing the subjects from reaching peak VO2 (from fatigue). 
However, despite them not being able to reach full potential exercise time, the HR and lactate were similar.

Finally, the bottom line-Time to exhaustion

The MLSS+10 caused a significant post interval decline in time x power.
This is certainly something we would like to avoid.  Imagine doing a great breakaway, only to have "nothing" left at the end.  

After reading the above study, I wanted to look into the effects of a MLSS+10 interval that was longer than the one I did for the lactate threshold testing.  I was hoping to see similar data as well as the post interval fatigue issue.  After a 1 hour warm up, I did an almost 11 min MLSS+10 capped off at the end with about 16 seconds max power.

  • The costal muscle O2 fell but did remain somewhat stable as did the RF and VL.  
  • Both the RF and VL were capable of a further drop at the end with the short max sprint.
  • The costal O2 did nosedive quickly from the 16 sec max effort.  This drop is much faster/deeper than it usually is, possibly because the respiratory muscles are already fully engaged and at their point of decompensation as well.  This was born out by the next figure, with the minute ventilation already approaching near max levels.
Now the respiratory data using the Hexoskin:

  • The heart rate remained in a stable range till the end (with the sprint).
  • Ventilation rose and continued to rise especially at the end with the max effort.

  • The breathing rate was about the highest I've had.  
  • Despite the very high rate, the ventilation was not at maximal until about 10 seconds later.  The respiratory rate had dropped back down somewhat but ventilation was actually higher.  Whether this is related to respiratory muscle fatigue (and hypoxia), inefficient shallow breathing is not clear but it is a pattern I have seen at the end of intense efforts.

Now for the post MLSS+10 "time to exhaustion" equivalent.  Instead of doing the study TTE protocol, I just did my usual 1 min max effort up the same hill I always do on my way home (for better comparison to previous efforts).  I usually average 510 to 540 watts, and that is after testing various intervals.

The 470 watt average was the lowest I've seen in many years.  The initial 10 second average power was almost 200 watts lower than my previous trek up the same hill last week. That 1 min max was still after doing a 5 min fast start intense interval (but not 11 minutes).
In short, the 11 min MLSS+10 resulted in severe deficits in performance later in the ride.
Here is a graphic of best ride average power this month vs today.  The 11 sec average is proportionately worse than the 1 min.

The costal O2 did drop nicely and I possibly could have pushed out a few more percent if I kept going another few seconds.  So despite the loss of peak and average power, the respiratory muscle hypoxia, cardiac output redistribution was still present, although possibly a bit less.

The Hexoskin data:
  • The max heart rate is near my usual peak (169 vs 171), the respiratory rate and ventilation both rise steadily with no "blips".  
  • The ventilation "efficiency" tracing looks great (did not have that paradoxical high rate but low ventilation volume) yet the average power was poor.  As in the above MLSS+10 study, the legs just did not have the ability to get to the point where respiratory fatigue really set in.  There certainly was a credible change in metrics, but not to the point where respiratory muscle fatigue became a major factor.

Takeaway lessons:
  • The difference in riding at MLSS vs MLSS+10w leads to several physiologic parameter changes.
  • Although HR, RF HHb/SmO2 show some small differences, they are unlikely to be of major help in a real time riding situation.  The VL does not differentiate between these zones very well.
  • Costal O2 is helpful in other interval and pacing strategies, but it does not seem to have the fine discrimination at MLSS+10, at least for 11 minutes.  It is possible curve shape would change at longer effort times.  However for this to be useful in pacing, we would want it to show evidence well before getting to the point of fatigue.
  • The minute ventilation is the parameter that can be measured in the field that best distinguishes the two zones (also seen with the lactate threshold test).  The steady rise of ventilation to near peak levels seems to herald the loss of steady state.  This makes sense as the disproportionate increase in ventilation is largely due to acidosis buffering.
  • In the post interval time to exhaustion bout, the ventilation actually dropped in the MLSS+10, probably related to the subjects just not able to push themselves long and hard enough.  In my example, the ventilation, respiratory rates rose with a post MLSS+10 effort.  However they did not show the fatigue patterns seen with optimally powered intervals.
  • The time to exhaustion is reduced by riding at just 10 watts above MLSS, with the practical implications easily realized.
  • In my case, a timed interval done with maximum power was reduced in both peak and average wattage.

Lactate related posts

Tuesday, August 14, 2018

Respiratory muscle training, benefits and discussion

Over the past year, I have focused on literature and personal observations about the behavior of some of the accessory muscles of respiration .  As exercise intensity rises, the competition between locomotor and respiratory muscles becomes more pronounced.  In fact, some studies have shown that unloading the respiratory system improves leg muscle performance.  Therefore, besides training the muscles of cycling or running one would guess that training the respiratory muscles would help overall endurance performance.  Indeed, many studies have shown that training the respiratory muscles (RMT) can help improve endurance exercise performance.  This post will be about both personal observations of using RMT as well as some selected literature discussion.  For an excellent comprehensive review see this recent publication.  

What are some reported mechanisms of RMT that lead to improving endurance performance?
  • Improvement in diaphragm muscle fiber types, strength and endurance
  • Reduction in respiratory muscle fatigue 
  • Reduction in the sense of breathlessness
  • Reduction in the respiratory muscle metabo-reflex (sympathetically induced vasoconstriction of limb blood flow at high ventilation rates)
  • Improved overall respiratory muscle efficiency
A word on the "metobo-reflex"
Early studies have shown that manipulating the work of breathing by loading/unloading the respiratory system can reduce blood flow to the leg.
Here is a figure from the paper above where inspiratory resistive loads were placed on subjects.  With a higher load, blood flow to the legs was reduced, with negative loads, blood flow was improved:

In addition this study was reviewed in my early post on respiratory muscle O2 monitoring.
Summing this up, in the competition for blood flow and O2 supply, the respiratory system components usually win.

Since we can't ride around connected to a CPAP machine or ventilator assist device, can training the respiratory system somehow achieve something comparable?  
I think we can....

The claims of RMT sound almost too good to be true, but the data is out there.   However, a certain frequency and quality of training is needed to reach and maintain the improvements.  In addition, there are different types of training devices that provide different varieties of physiologic challenges and choosing the correct device is important.
An excellent study compared the EMG activity of the multiple muscles involved with 3 different devices.

Looking at the devices (descriptions from the paper):

Voluntary isocapnic hyperpnea
VIH was conducted using SpiroTiger M 1.4 (idiag AG, Fehraltorf,
Switzerland). The set up was individualized according to the manufacturers
recommendations: bag volume according to 50% of vital capacity;
minute volume was set as 60% of MVV; target tidal volume was
set as bag volume × 1.2; target breathing frequency was set as minute
volume · target tidal volume−1. Two exemplary bouts of VIH lasting
1 min with a 30 s break in between were applied 

Inspiratory flow resistive loading
IFRL was conducted using RespiFit in the endurance mode setting
(Biegler, Mauerbach, Austria). The set up was individualized according
to the manufacturers recommendations: tidal volume was assessed
while breathing through the device with 14 breath/min for 30 s set at
80% of the individual PImax. Two exemplary bouts of IFRL lasting
1 min with a 30 s break in between were applied.

Inspiratory pressure threshold loading
IPTL was conducted using POWERbreathe (IPTL-PB) (Powerbreathe
Classic, HaB International Ltd, Warwickshire, UK), set at 80% of the
individual PImax based on recommendations for IPTL (Janssens et al.,
2013). Furthermore, a variant of IPTL was conducted using RespiFit in
strength mode with identical settings (IPTL-RF). Target pressure was
assured with a separate pressure transducer (see below). Two bouts of
each 5 inspiratory efforts with a 30 s break in between were applied
with each device.

The results were as follows:

As noted the Powerbreathe (and Respifit in strength mode) lead to the best diaphragmatic activation.  Their conclusion:

From the standpoint of which device to use, the above is helpful, but some further refinement in technique can produce even better diaphragm training.
According to this study, using an abdominal breathing technique, diaphragmatic activation is markedly enhanced.  From the paper:
"Trial 2 was identical to trial 1,
except for the breathing instructions. Participants were
instructed to emphasize the use of their diaphragm and ensure
their abdomen ‘‘sticks out’’ during each inspiratory
maneuver (diaphragmatic inspiratorymuscle training [IMTdi]).
Subjects were familiarized with diaphragmatic breathing by
sitting upright in a chair and placing one hand on the abdomen
and the other hand on the ribs along the anterior
axillary line. Subjects would then breathe and attempt to
keep their hand on the rib stationary and only moving their

And the EMG responses:

Whether this translates to better net cycling/running performance has not been shown.

What I use:
I started using the Powerbreathe K4 almost 1 year ago.  Using the manufacturers recommended protocol, the resistance level gradually rose over several weeks and I informally noted better 1 min, 3 min interval times.  However, after about 3 mos, I slacked off due to some cervical spine issues and laziness.  I did pick it back up again at about the same time as I began using the Hexoskin.  I have some interesting observations to discuss and perhaps some tips on usage.

Let's look at what happens to the costal O2 during a 30 breath (about 3.5 minutes) session:

There is a definite per breath ripple and gradual reduction in costal O2.  Interestingly, the dip at the end was after stopping, which is also seen in some of the post interval tracings.
The Total Hb also shows a ripple on each breath potentially showing either a flow change or effect from muscle compression:

Before getting too excited about the training effect related to SmO2, the R RF tracing also shows a per breath spike.  I think this is more from "tensing" the legs during each maximal breathing effort:
But it does not have the post RMT drop in O2 that the costal muscles exhibited. 

Sample of Data from the Powerbreathe K4:

One of the perks of this particular model is the ability to look at both per breath dynamics as well as your progress over time.  It was due to this functionality that I stumbled upon the observation that my power/flow readings were much better after a ride than before.  There have been papers discussing the potential benefits of doing RMT while exercising and perhaps this is somewhat related.
For example, here is a 30 breath graphic from before a 3 hour ride:

And after riding 3 hrs (with several intense intervals):
Both volume, power, flow and energy are all much higher when done immediately after the ride (no more than a couple of minutes separating).  I have also noted that delaying by even 30 minutes will largely eliminate this effect.
I would imagine it is superior to do it this way as you will be achieving much better physiologic stimulation of the various training goals.  For example, lifting a heavier load and more often in a weight training session will provide a better stimulus for muscle development.  I have basically stopped doing the RMT randomly and now only use it immediately after a ride, 5 days a week.

Another example of the Powerbreathe data analysis is the breath by breath analysis:
Each parameter is graphed over time giving clues on improvements in flow, volume and power.

Back to the pre/post ride comparisons
Initial raw thoracic and abdominal breathing:
First tracing is before a 3 hour ride and represents one breath in the middle of the 30 breath session, purple is thoracic and blue is the diaphragmatic estimation using the Hexoskin.
 Next is the same raw view after the 3 hour ride (one breath):

The other breaths are basically the same and show much better motion, especially for the abdominal component after the ride.  
The difference in the blue tracing is striking and indicates better net diaphragmatic breathing motion if training immediately post ride.

What about performance?

It's great that the Powerbreathe resistance numbers improve with training but does that actually translate to one's personal muscular improvement?  For instance, do we get better diaphragmatic strength or excursion?  I looked at the Hexoskin raw data from the end of first week of use (1 week into usage was the first post ride data), to present (about 3-4 weeks)
After 3 weeks of RMT the one breath curve (post ride) looks like this: 

On first look, I wasn't sure there was much difference.  However, the time scale is about the same and the diaphragm motion pattern seems deeper in the first few seconds.
Whether this represents a training effect is unclear.  The curve shape in both pre and post RMT were the consistent over the 30 breaths.
Bottom line- possible improvement in diaphragmatic excursion during post ride training and net improvement as well over 3 weeks.

Breathing rate and minute ventilation

Here is my first 1 minute max interval done with the Hexoskin that was near the beginning of Powerbreathe training:
I circled a spike in respiratory rate that corresponded to a time of very severe breathlessness, but not a higher ventilation volume-more on that below.

The power, costal O2 corresponding:
The BSX sensor is on my L which is it's new usual location (because of the Hexoskin shirt dongle placement on the R).  

Now for the more recent post RMT tracings:
No late spike in rate of breathing.  Steady rise in minute ventilation.

And Costal O2:
  • The net average power was better in the post RMT.  I have done many of these 1 min max intervals (at least 1-2 a week).  During the hot summers here, power generally drops from a winter average of about 530 to about 510 watts.  It is highly unusual for me to get the better figure in the heat and earlier in the week I reached 545 watts on the same hill.  The last time I was able to do this in warm weather was last summer after doing the RMT training 2x a day for 6 weeks.
  • The respiratory rate rose briskly in both pre and post RMT intervals but there was a later period of very high rate in the pre RMT session.  I was aware of being extremely breathless at that point and decided to pay more attention to the Powerbreathe training as a result.  Despite that elevation in rate, the minute ventilation was not better than my usual maximum.  Could this represent a less efficient state, muscular weakness, etc?  I definitely experienced severe breathlessness.  Although I was certainly winded in the "post RMT" interval, it was not accompanied by severe breathlessness.
  • Despite a higher average power on the post RMT interval, the costal desaturation was less, and the recovery faster.  This was seen in a similar interval at 545 watts during that week.  Although placement may effect nadir measurements, it is also possible that the RMT has changed the costal desaturation by either improving efficiency, improving the diaphragmatic part of breathing and "unloading" the accessory muscles somewhat.
  • The max heart rate was about the same in each.

Finally to pull the concepts together we have this excellent paper:

And this:

With the conclusion:

To summarize:
  • The respiratory muscles compete for blood supply with the locomotor related musculature especially during high intensity exercise.  Under most circumstances the respiratory system will win.
  • Training both systems has been shown to improve performance.  My guess is that most cyclists, runners and skiers neglect the respiratory component.  Therefore RMT is a potentially helpful technique for most endurance athletes.
  • The type of RMT, as well as the conditions of training may affect outcome improvements.
  • If engaging in RMT, keep track of some benchmark performance measurements to see the magnitude of change (if any).  There will probably be a range of "responders" and non responders.
  • Research indicates that the change in abdominal pressure and diaphragmatic motion may change blood partitioning to optimize performance.  Taking advantage of this phenomenon by RMT seems possible and another reason to employ it.