Ionic's tracking of heart rate

[marcwilliams25]

My first Fitbit was the Surge, bought it for the GPS more than anything (track my hikes) I was so impressed by it that this year I thought I'd upgrade to the Ionic, got to admit that now it seems like a poor decision. The Surge seemed to track my heart rate during workouts quite accurately. It picked up on peaks where I'd push myself during a session. The Ionic is nowhere near as accurate in my opinion.

To give you an example, yesterday I just did a gentle 30 minute run on a treadmill, nothing really that taxing. When I synced my Fitbit apparently, my heart rate was in the peak zone throughout the 30 minutes. Today I did a "3 rounds for time" workout that had a 20 calorie assault bike as an element, after I finished it I couldn't get off the floor for about five minutes, I was totally gassed, when I synced my Ionic my heart rate was just seven minutes in the cardio zone. That can't be accurate

Moderator edit: Updated subject for clarity 

[tomp2]

First off, I'd be interested to know what you're using to extract line graphs with that much detail? Thanks.

I had a very interesting session yesterday. I try to do a session on the Wattbike at my local gym once a week at this time of year. I did one of its very mean HIIT sessions yesterday, involving 15 hard efforts in a 24 minute session, split into three groups of five. The efforts are each 20 seconds and have about 40 seconds recovery between each. This is what the session looked like after completion, from the Wattbike app on my phone. The feed for this comes from my Polar HR strap to the bike, which then feeds the app, which collects the data in pretty good detail. Can be downloaded and analysed further:

Wattbike 24 10 18

 Cadence and wattage peaks are clear. Unfortunately the graph's scale, and the fact that I can do this exercise without too much cardio trauma, means the HR trace isn't as revealing. However, HR usually went from about 130 to 148 in each peak.

I also wore the Ionic, of course. I found while warming up, that keeping the back of my hand in line with my forearm was giving me almost identical readings to the Wattbike app figures from the Polar strap. The Wattbike has road bike-like dropped handlebars, complete with fake brake hoods. Riding with hands on the hoods flexed my wrist, and readings on the Ionic then became erratic. They settled down again with a straighter wrist. I didn't think the flexing was that extreme, but apparently the watch thought differently. Wait, though! There's more.

I was seeing my HR numbers do what I expected on the app - pretty much an instant pick up until the peak effort was over, then a satisfyingly sharp drop back as the rest phase began. I glanced frequently at the Ionic too, and was shocked/surprised to see it almost doing the reverse! As I put the effort in, its figure (reading from the standard "Exercise" mode) FELL in each peak, and in the recovery phase, rose back pretty quickly - though never to the kind of peak the HR strap was showing - before falling back to a resting figure that more or less matched the Polar strap number. I could affect the rate of recovery the Ionic showed if I dropped my left hand down loose. It generally "peaked" faster and recovered slower. I didn't try achieving peak power/cadence with one hand hanging by my side. The effort needed is too intense to allow that without almost falling off the machine.

Now, as others have pointed out, this reinforces the view that the watch and the strap are simply recording different things. The HR strap electrodes are catching heart beats. The watch is catching the vibrations from blood flow quite a bit further down the line, on the back of my wrist. Sure, simple. But during full-on effort, I'd be certain the body prioritises where peak blood flow goes - ie into the cardio-vascular organs. The back of the wrist doesn't feature as one of these, and if it therefore actually receives a DROP in blood flow, I'd not be surprised. I thought of the number of distance runners I know who have to wear gloves, even in summer, while racing, because their hands get so cold. 

Peak effort over, the blood supply around the body presumably levels out quite fast. Faster if I let gravity give it some help by taking my left hand off the bars and dropping my left arm to my side.

Result was that any graph from the Ionic will as like as not only show a reduced "echo" of the peak efforts. I need to do some reading up on more scientific comparisons between body and wrist-based systems.

Tom

[lenny4d]

@tomp2"The efforts are each 20 seconds and have about 40 seconds recovery between each" - I doubt you can see anything like that type of resolution in the fitbit HR exercise graph, simply because it is heavily filtered.

[SunsetRunner]

@tomp2 The graph is actually useless when it comes to detail analysis. Fitbit doesn't support HR export but there is a trick. To get exact data from Ionic you log your exercise always using activity with GPS. For example, walk, run. or golf Mind that first two may probably trash your stride if you use automatic calculation one ( and if you care about stride, I don't 😉 ). I am not sure about Golf but I assume this exercise shouldn't use stride at all for anything. I use Golf anyway. Then you can export TCX files which contains HR second after second. Then I run one of my scripts to extract the data I need:

So I have a timestamp and BPM ( I just recorded that for about 2-3 minutes ). The same I do with Polar chest strap. PolarBeat synchronizes with PolarFlow from which I can export exact HR. In case of Polar you don't need to fake workout as non-GPS activities also produce TCX files. I have no idea why Fitbit found it impossible ( well, among many other things ). I should have mentioned that in the beginning but this is the reason for using Golf ( or any other activity with GPS ). Otherwise, TCX file is not produced.

With such data, you can do literally everything. I wrote tools that can extract and analyze anything coming from TCX ( like GPS data, distances, pace etc. ). To make custom comparison graphs I use gnuplot tool. When you have actual data there's nothing really to hide.

After extracting the activity you can manually log non-GPS one using details from Golf ( calories and other things ) and remove Golf. This way you will keep your log clean.

[tomp2]

Clever idea. My computer skills are not up to that, unfortunately.

Tom

[SantiR]

Hello everyone! Thanks for your continuous participation on this thread regarding Ionic's tracking of hear rate data. Thanks for sharing all of your experience here with screen shots and helpful feedback.

In this case, what helps the most to get an accurate heart rate recording is the wear the watch accordingly to the type/intensity of activity you are performing. 

When not exercising try using this one:

When exercising you can experiment with wearing the device higher on your wrist during exercise for an improved fit and more accurate heart-rate reading. Many exercises such as bike riding or weight lifting cause you to bend your wrist frequently, which could interfere with the heart-rate signal if the watch is lower on your wrist. In such cases, try using this one:

Also, make sure your watch is not too tight, but not too loose on your wrist. Please check more details here.

Hope this helps. Keep me posted! 

[tomp2]

Thanks Santi,

However, I don't think this moves us forward. I bought a "Sports" strap (the one with lots of holes in it) for my Ionic specifically because it has more levels of adjustment than the heavier one that came with the watch. I do a range of different exercises in the gym, and adjust it as appropriate to the likelihood of poor readings on account of wrist flexion etc. However, my experience (even though I am quite a new user) is that, although this might make a small difference when wrist flexion is involved, it makes little, if any difference when the exericse involves swinging one's arms. 

Yesterday was the first time I ran (treadmill) with my Ionic. I've had a foot injury that has prevented this up to now. I ran wearing a Polar HR strap too. This communicates with the treadmill readout and was done in order to cross-check the figures from the Ionic. I was already warmed up, and seeing around 90 bpm on the Ionic before I started. Normal for me. Immediately I started running (only at 8kph), my Ionic readings shot up to 135 and stayed at that level. By comparison, the Polar strap showed a slow rise to about 115bpm, then slowly to about 125. Again, the latter is normal, not the Ionic reading.

I thought I may have the Ionic too loose, and adjusted it up my wrist a little. This made no difference whatsoever to the readings. However, If I jogged with my arms hanging by my side, I got readings from the Ionic that were far closer to the Polar strap readings. This reinforced my opinion that arm movement is one factor in overstating the Ionic readings.

I finished my run, and moved to an elliptical cross-trainer machine that has stationary handles as well as "swing" handles. I removed the Polar HR strap, because sensors on the cross-trainer handles give what I have always found to be a reasonably accurate heart rate reading. I began my session using the stationary handles. My wrists were not flexed very much when using these. When I began exercising, I got Ionic readings that were quite closely in line with those of the elliptical trainer. The trainer is, of course, reading pulse through hands and fingers gripping the sensors. 

I then switched to the "swing" handles for a more intensive phase of my workout. I would expect my HR to rise at this point. Indeed it did on the cross-trainer sensors, to levels I am accustomed to in this kind of exercise. However, the Ionic figures rose very quickly, to levels way beyond what I would expect. Again, I wondered whether I had the Ionic in the wrong position, as my wrists were now flexing more than before. In adjusted the Ionic a little, but it made no difference. My conclusion again is that it is the movement of the watch itself, whether tightly strapped or loosely, and regardless of the precise position on the wrist, that was causing the misreadings.

Tom

[Freeco]

I tried the trick with the wrist band covering the Ionic twice now. It does seem to improve HR measurement a bit. I still see drops in the chart where I'd expect a more steady HR.

Downside is that my wrist gets completely soaked in sweat 😉 And I can't easily check the screen while exercising anymore.

[Georgiy]

That is a nice research. I am thinking about upgrading my Charge 2 to Ionic. As far as I understood the latter is supposed to count steps on elliptical more precisely. However, if it messes up HR and corresponding calorie burn, it is definitely deal-breaker for me.

I mean you can export data and run python script and stuff and then manually add the data to the app and eventually end up with what I am doing for my Charge 2. With one difference though: I am adding distance/steps :).

Regarding the scripts, it seems that there is a way to get non GPS HR data with 1 second resolution via their web API. Or to get real-time data via device API. Both seem to be kinda overkill (to say the least). However you can check it easily with something like Postman I believe.

Here is the link to web API stuff.

Unfortunately, 1 second resolution does not mean that you will actually get it. I just tried web API to get my elliptical HR, and it is between 1 and 3 seconds as you may see.

{
    "activities-heart": [
        {
            "customHeartRateZones": [],
            "dateTime": "today",
            "heartRateZones": [
                {
                    "caloriesOut": 0,
                    "max": 93,
                    "min": 30,
                    "minutes": 0,
                    "name": "Out of Range"
                },
                {
                    "caloriesOut": 11.59438,
                    "max": 130,
                    "min": 93,
                    "minutes": 2,
                    "name": "Fat Burn"
                },
                {
                    "caloriesOut": 0,
                    "max": 158,
                    "min": 130,
                    "minutes": 0,
                    "name": "Cardio"
                },
                {
                    "caloriesOut": 0,
                    "max": 220,
                    "min": 158,
                    "minutes": 0,
                    "name": "Peak"
                }
            ],
            "value": "102.41"
        }
    ],
    "activities-heart-intraday": {
        "dataset": [
            {
                "time": "09:30:01",
                "value": 98
            },
            {
                "time": "09:30:04",
                "value": 98
            },
            {
                "time": "09:30:07",
                "value": 98
            },
            {
                "time": "09:30:09",
                "value": 99
            },
            {
                "time": "09:30:11",
                "value": 100
            },
            {
                "time": "09:30:12",
                "value": 100
            },
            {
                "time": "09:30:13",
                "value": 101
            },
            {
                "time": "09:30:15",
                "value": 103
            },
            {
                "time": "09:30:18",
                "value": 103
            },
            {
                "time": "09:30:21",
                "value": 103
            },
            {
                "time": "09:30:24",
                "value": 102
            },
            {
                "time": "09:30:27",
                "value": 102
            },
            {
                "time": "09:30:28",
                "value": 101
            },
            {
                "time": "09:30:31",
                "value": 101
            },
            {
                "time": "09:30:34",
                "value": 102
            },
            {
                "time": "09:30:35",
                "value": 103
            },
            {
                "time": "09:30:37",
                "value": 104
            },
            {
                "time": "09:30:38",
                "value": 105
            },
            {
                "time": "09:30:41",
                "value": 105
            },
            {
                "time": "09:30:42",
                "value": 104
            },
            {
                "time": "09:30:45",
                "value": 103
            },
            {
                "time": "09:30:48",
                "value": 103
            },
            {
                "time": "09:30:50",
                "value": 104
            },
            {
                "time": "09:30:52",
                "value": 105
            },
            {
                "time": "09:30:55",
                "value": 105
            },
            {
                "time": "09:30:58",
                "value": 106
            },
            {
                "time": "09:30:59",
                "value": 107
            }
        ],
        "datasetInterval": 1,
        "datasetType": "second"
    }
}

TCX is supposed to be created regardless of presence of GPS data according to what is returned via API. Query is something like

https://api.fitbit.com/1/user/-/activities/list.json?afterDate=2018-11-08T20:30:00&limit=20&offset=0&sort=asc

And it returns something like

            "originalDuration": 1200000,
            "originalStartTime": "2018-11-08T20:37:05.000+01:00",
            "source": {
                "id": "22D54K",
                "name": "test",
                "type": "app",
                "url": "https://127.127.127:8088"
            },
            "speed": 0.08220000000000001,
            "startTime": "2018-11-08T20:37:05.000+01:00",
            "steps": 37,
            "tcxLink": "https://api.fitbit.com/1/user/-/activities/17994492611.tcx"

I have no slightest clue how distance and number of steps are related when you are manually adding elliptical activity. It does not make sense to me. Even if you can add distance via API and it passes validation, you will not see correct values or either.

I may understand why they have removed option to edit activities that contain distance - it is all about trophies and achievements. However there is no option if you don't care about that cr@p and just want to have correct data for yourself. Otherwise not only something may not work as you would expect, but also nothing is done (or more - this functionality is removed) to help you at least enter correct values yourself.

[SunsetRunner]

@Georgiy I think the HR you get through the API is the same you can see on the graph in the dashboard. It also misses certain timestamps. The TCX file, however, contains all the details except when HR cannot be read ( which in my case happens, too ). I already gave up on Ionic. I stopped tracking my activities with an exception of automatically tracked walking ( everything else I turned off as I was getting things like "bike" when driving through slow traffic etc. ). It's just pointless. Whatever we show, whatever evidence we put on a table it all falls on deaf ears. It doesn't get anywhere beyond this forum. Unfortunately for me, I can't afford the next fitness tracker right now due to other more important expenses so I'm stuck with this piece of garbage for at least two months more.

[Freeco]

---

@Georgiy wrote:

I am thinking about upgrading my Charge 2 to Ionic. As far as I understood the latter is supposed to count steps on elliptical more precisely.

---

If that's your only reason to change to an Ionic, I'd keep my money in my pocket.
I've used a Surge for 3 years in combination with elliptical workouts. It's step count was only 1/3.
I've changed to an Ionic about 5 months ago. It's step count during elliptical workouts is only slightly better, 40-45% perhaps, while the HR measurement is worse.

[lenny4d]

@Georgiy

>> I am thinking about upgrading my Charge 2 to Ionic. As far as I understood the latter is supposed to count steps on elliptical more precisely.

@Freeco

>> I've used a Surge for 3 years in combination with elliptical workouts. It's step count was only 1/3.
I've changed to an Ionic about 5 months ago. It's step count during elliptical workouts is only slightly better, 40-45% perhaps, while the HR measurement is worse.

I have noticed threads where ppl complain that steps are tracked on the elliptical at all, suggesting it should be zero! And similarly for rowing machines and whatever not. It's like "steps are steps, and nothing else." It's a fundamental question. 

For me HR on the treadmill is spot on to around 140-150, but on the elliptical it stops around 120, and at 130 it is 30 less than reality, which makes it useless if not dangerous. Fitbit should have offered a blue-tooth connected HR monitor (chest or otherwise) for improved HR accuracy during hi-intensity workouts. Simple as that.

[SunsetRunner]

@lenny4d I had exactly same experience on the elliptical and Ionic. Ionic always measure way lower HR than my real HR. The difference is 20-30bpm. Sometimes more. It's a lot. Using Ionic on the elliptical is pointless in my case as all HR data is just rubbish.

[Freeco]

---

@lenny4d wrote:

I have noticed threads where ppl complain that steps are tracked on the elliptical at all, suggesting it should be zero! It's like "steps are steps, and nothing else." It's a fundamental question.

---

we're going off-topic whit this, but indeed, I know some people don't agree with counting steps on an elliptical. I'm not one of them. It's the same leg movement as walking/running, just without the shock on your joints when you put your foot back on the ground. With the same logic running on a treadmill wouldn't count as running either, as you're not really moving forward.

Back on topic: in the meantime I've bought a new chest strap to replace the cheap ones from Decathlon I've been using the past few years. A simple Polar T31, which can connect to my elliptical and works with its workout programs.
One of the following days I'll do the cardio program again setting the target HR to eg a steady 150 bpm. I'll see how the Polar performs, and how it compares to my Ionic.

[Georgiy]

@Freeco I second this. Step count must work properly.

And poor HR measurement by Ionic is surprise for me actually. 

Charge 2 in this regard is enough. I checked it against chest strap and the measurement accuracy is satisfactory.

I dunno how they managed to mess up this on a new and supposedly improved hardware. The only reason that comes to my mind is that there was serious pressure from management side to release product earlier that led to poor testing. Pretty much the same case as with 6th generation of Samsung Galaxy devices.

[SantiR]

Hello everyone! Thanks for your continuous participation on this thread regarding the heart rate tracking on Ionic. Thank you all for posting your research and conclusions about this. I'm sure this will be very helpful for other users who might be experiencing something similar.

If you keep experiencing trouble with tracking; please check this post for some information about this.

Keep me posted! 

[tomp2]

Another two weeks of usage and testing of my Ionic against other heart rate devices is leading me to be extremely sceptical of the Ionic’s value to anyone who takes their training seriously. 

I’ve tested the device on a spinning bike, a Wattbike, and several kinds of elliptical crosstrainer while waiting for a foot injury to heal. Now it’s getting better, I am doing a little slow running on a treadmill at my gym. Every time I start, my heart rate readings on the ionic simply go through the roof for the first few minutes. I have a properly measured HR max of 160 bpm, and I am familiar with training at that kind of level. Today, the Ionic hit 182 bpm within a minute, while I was still barely jogging. It stayed there for several minutes. Nothing wrong with how I was wearing the thing: a few minutes later, my recorded heart rate was back down to more credible levels.

It’s becoming my view that the algorithm that runs the Ionic has trouble coping properly with arm movement. 

Tom

Moderator Edit: Format

[SantiR]

@tomp2 Thanks for your reply and for sharing those details about the results of your testings. 

So it tracked an inaccurate heart rate reading when you first started the exercise and then it recorded accurate readings? Very interesting. Thanks for sharing this information.

Keep me posted! 

[lenny4d]

@SantiR@tomp2 

One possibility for those jumps is that the reading you get is PULSE + CADENCE (bpm + steps_per_min)! The software is supposed to filter out the step induce noise but obviously that can be challenging -- because cadence and bpm are often enough of the same magnitude order - and potentially the algorithms get confused. I bet my $ on it that that is what you are seeing.

Also look at the following article on how the HW really works under the hood, comparing fitbit, Garmin and other brands, very enlightening for those technically minded. And scary if you want to know the truth....  

https://medium.com/@pkuhar/comparing-optical-heart-rate-monitor-technical-implementations-450c6b35bd...

(edited link, something had gone wrong)

The big difference between the (imho) very robust and accurate Charge 2 and the new trackers like Versa and Ionic may very well be in the frequency at which they pulse the LEDs, and/or the amount of processing that is done, or lack thereof -- to save power.  Here is also some useful info: http://fellrnr.com/wiki/Optical_Heart_Rate_Monitoring  Note section 4.

[tomp2]

Thanks. These look interesting.

I've been sure that what I and others are seeing was actually a better-understood problem than the manufacturers are prepared (for obvious reasons) to admit to customers who have just shelled out good money on something that is not nearly as sophisticated as the publicity claims. I'm also sure that the very frequent suggestions that a re-start of ones watch will cure all manner of ills is nothing more than a poor effort to conceal this. It will not alter fundamental shortcomings in the hardware and basic algorithms.

Of course, I have only come to realise this after purchasing my Ionic. I find it a useful tool for a few basic tracking functions, like sleep and general  movement. However, I was unfortunately taken in by reviews that claimed it was a good sports watch, and I am not finding that to be true.

[tomp2]

I've just clicked on the first of the links in your post. It seems to have gone. Second one is still there.