Heart rate modelling as a potential physical fitness assessment for - - PowerPoint PPT Presentation

▶
heart rate modelling as a potential physical fitness
SMART_READER_LITE
LIVE PREVIEW

Heart rate modelling as a potential physical fitness assessment for - - PowerPoint PPT Presentation

Nov 14, 2023 27 likes •308 views

Heart rate modelling as a potential physical fitness assessment for runners and cyclists Dimitri de Smet , Marc Francaux, Julien M. Hendrickx and Michel Verleysen Universit catholique de Louvain Belgium Machine Learning and Data Mining for

slide-1
SLIDE 1

Heart rate modelling as a potential physical fitness assessment for runners and cyclists

Dimitri de Smet, Marc Francaux, Julien M. Hendrickx and Michel Verleysen Université catholique de Louvain Belgium Machine Learning and Data Mining for Sports Analytics Workshop at ECML & PKDD 19th September 2016

slide-2
SLIDE 2

Outline

  • Context and machine learning perspectives
  • Contribution
  • Novel heart rate parametric model
  • Parameters identification
  • Validation
  • Conclusion

1

slide-3
SLIDE 3

Context : automated coaching

Coaches are more and more relying on scientific approach that requires a model

2

Athlete Adaptation Model Workout Sessions w(t) Performance p(t)

slide-4
SLIDE 4

3

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

Exogenous variables

Context : performance

slide-5
SLIDE 5

4

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

  • 5km running time
  • 100km running time
  • 100km cycling
  • …

Exogenous variables

Context : performance

slide-6
SLIDE 6

5

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

  • Wind speed
  • Temperature
  • Altitude
  • …

Exogenous variables

Context : performance

  • 5km running time
  • 100km running time
  • 100km cycling
  • …
slide-7
SLIDE 7

6

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

  • Athlete state
  • Endurance
  • Power
  • Flexibility
  • Skills
  • …
  • Wind speed
  • Temperature
  • Altitude
  • …

Exogenous variables

Context : performance

  • 5km running time
  • 100km running time
  • 100km cycling
  • …
slide-8
SLIDE 8

7

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

  • Quantification

based on :

  • Duration
  • Distance
  • Heart rate

zones

  • …
  • Athlete state
  • Endurance
  • Power
  • Flexibility
  • Skills
  • …
  • Wind speed
  • Temperature
  • Altitude
  • …

Exogenous variables

Context : performance

  • 5km running time
  • 100km running time
  • 100km cycling
  • …
slide-9
SLIDE 9

Machine learning perspectives

8

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

Model building and athlete fitting require input-

  • utput examples:
  • Inputs (workout sessions)
  • Output (performance)
slide-10
SLIDE 10

Data profusion

  • Many activity sharing platforms
  • Activity export = Track

9

Track Name : Morning ride Timestamp Location (lat, lon) Altitude Heart Rate Power Output … 2016-06-21 13:27:28.00 (50.4307, 3.736080) 16.60 m 97 bpm 125 W … 2016-06-21 13:27:29.00 (50.4308, 3.736082) 16.62 m 96 bpm 147 W … … … … … … …

slide-11
SLIDE 11

Data profusion

  • Track content

10

slide-12
SLIDE 12

Data profusion

  • Track content

11

slide-13
SLIDE 13

Data profusion

12

(strava heatmap)

slide-14
SLIDE 14

Machine learning perspectives

13

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction

Model building and athlete fitting require input-

  • utput examples:
  • Inputs (workout sessions) : massively available
  • Output (performance) : sparse
slide-15
SLIDE 15

Co Contri ribution

14

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction Heart Rate Parameters Fitting Heart Rate hr(t) Power Output po(t) Heart rate parameters

slide-16
SLIDE 16

Co Contri ribution

15

Athlete Adaptation Model Workout Sessions w(t) Performance p(t) Sport

… …

Features extraction Heart Rate Parameters Fitting Heart Rate hr(t) Power Output po(t) Heart rate parameters

contribution

slide-17
SLIDE 17

Heart rate modelling

16

Workout Session w(t) Heart Rate Parameters Heart Rate Parameters Fitting Heart Rate hr(t) Power Output po(t) Athlete’s heart rate parameters Parametric Heart Rate Model Heart Rate hr(t) Power Output po(t)

slide-18
SLIDE 18

Heart rate model : steady state

17

Parameters

  • HR max [bpm]
  • Resting HR [bpm]
  • Slope [bpm/w]
slide-19
SLIDE 19

Heart rate model : steady state

18

Parameters

  • HR max [bpm]
  • Resting HR [bpm]
  • Slope [bpm/w]
slide-20
SLIDE 20

Heart rate model : transient response

19

(measurement)

d hr(t) dt + 1 τ hr(t) = po(t)

hr(t) = hr(t0) + (hrss(po(t)) − hr(t0))e− t

τ

HR(t+1) = ( HR(t) + 1

τr (HRss(po(t)) − HR(t)),

if HRss(po(t)) ≥ HR(t) HR(t) + 1

τf (HRss(po(t)) − HR(t)),

if HRss(po(t)) < HR(t)

slide-21
SLIDE 21

Heart rate model : fatigue

20

  • Intra-session workload results in fatigue that

induces increased heart rate for the same power

  • utput
  • Modeled by replacing po(t) by

po(t) + kf Z t

t0

po(t)dt

slide-22
SLIDE 22

Heart rate modelling

21

Athlete’s heart rate parameters

  • Resting HR [bpm]
  • HR slope [bpm/watt]
  • HR max [bpm]
  • Rising time constant [s]
  • Falling time constant [s]
  • Sensitivity to fatigue

[Watt/Joule]

Parametric Heart Rate Model Heart Rate hr(t) Power Output po(t)

slide-23
SLIDE 23

Cycling activities results

22

slide-24
SLIDE 24

Energy cost of running

23

Minetti, A. E., Moia, C., Roi, G. S., Susta, D., & Ferretti, G. (2002). Energy cost of walking and running at extreme uphill and downhill slopes. Journal of applied physiology, 93(3), 1039-1046.

slide-25
SLIDE 25

Running activities results

24

slide-26
SLIDE 26

Validation

  • 72 Cycling activities (3 cyclists) average rmse : 4 bpm
  • 234 Running activities (2 runners) average rmse : 6 bpm

25

Cardiac parameters Heart Rate Model Heart Rate hr(t) Power Output po(t)

slide-27
SLIDE 27

Conclusion

  • Pressing need for continuous fitness assessment
  • Our identified parameters allow for accurate heart

rate simulation

  • Ongoing research
  • Our parameters Vs Laboratory measurements
  • Performance prediction (like race times)

26

slide-28
SLIDE 28

Questions ?

27