# DecreasedRmssdAlgorithm

The goal of this algorithm is to detect when psychological stress causes a derease in the heart reate variability parameter RMSSD and then trigger a questionnaire. The algorithm is based on Verkuil et al. 2016 (opens new window).

# Sensor requirements

To use the Decreased Rmssd Algorithm, the ECG and Activity sensor EcgMove 3 or EcgMove 4 is required.

# Algorithm description

Heart Rate Variability (HRV) is associated with physiological and psychological wellbeing of a person. Decreased HRV corresponds to physical and psychological stress, and baseline HRV indicates overall physical fitness. When engaged in intense physical exercise, HRV value decreases, and HRV can also decrease due to psychological stress. The goal with this algorithm is to detect when psychological stress causes this decreased value, and then trigger a questionnaire. HRV is measured and assessed using RMSSD (which is the square root of the mean of the squares of the successive differences in-between adjacent heart beats).

To sum up, this algorithm aims to increase the probability to trigger a questionnaire based on a psychologically caused decrease in RMSSD. It’s been used for studies assessing psychological stress isolated from stress related to physical activity. However, this algorithm also has some limitations.

  • Firstly, to use this algorithm effectively, a pre-study must occur to capture the relevant parameters for each participant. (More on this point later.)
  • Secondly, despite this data, the sensor can only isolate and remove the exertion effects from the body when it’s moving through space. The decreased state of RMSSD from any stationary exertion such as stationary cycling or lifting heavy objects will not be isolated and the corresponding period can be mistakenly interpreted as psychological stress.

Utilizing a logic chain within movisensXS to determine whether any physical exertion occurred at the time of the alarm can counteract this problem. Simply asking if the participant experienced physical exertion as the first question, and then making the following questions visible when “No” is selected achieves this goal.

Decreased Rmssd Algorithm Overview

The algorithm tracks periods of Decreased RMSSD activity, and then assesses and isolates the physical causes utilizing the following steps. When the algorithm concludes the cause of Low RMSSD is not from physical activity, a trigger is sent to movisensXS.

  1. The data inputs for the algorithm are:

    • HRV RMSSD [ms], sample rate 1/min
    • HrvIsValid, sample rate 1/min
    • Movement Acceleration [g], sample rate 1/min
  2. It calculates the predicted average RMSSD level based on the maximum level of movement measured during the previous “Window length of Movement Acceleration [min]”.

    1. This continuous prediction of RMSSD can take the form of one of these function:

      • Predicted RMSSD = intercept + slope x movement (linear relation)
      • Predicted RMSSD = intercept + slope / movement (inverse relation)
    2. Values for regression method, intercept and slope for each individual have to be entered for each participant during coupling the study, using calibration data acquired during resting and exercise.

    3. A RMSSD threshold can be defined, as a measurement error has to be taken into account. If the actual RMSSD falls below the Predicted RMSSD minus the threshold this minute is tagged as a “Period of Low RMSSD”.

      • RMSSD for comparison = Predicted RMSSD - Threshold
      • Threshold has to be configured for each participant during coupling the study
  3. The algorithm keeps track of the number of periods (minutes) where the RMSSD is below the configured threshold. To do this, you configure the RMSSD Assessment Window - a window of time (in minutes) - and allocate the number of 1 minute periods in which the RMSSD should be below the threshold. For instance, over a 10 minute Window you may require 8 periods (minutes) of Low RMSSD – that is, RMSSD that is under the configured threshold. The Length of Window and the amount of Periods can be configured.

  4. When the configured number of periods in Low RMSSD has been detected, a trigger is sent to movisensXS.

    1. The total amount of prompts indicating prolonged periods of low RMSSD per day must be configured in movisensXS
    2. The prompts indicating prolonged periods of Low RMSSD should be interspersed with randomly triggered ‘neutral prompts’ at which no “Low RMSSD” is present during the preceding 10 minutes.
    3. Minimum amount of time between the prompts is: 60 minutes.
  5. When no valid RMSSD values are available then no trigger will be provided (HRV data not valid = No period of Low RMSSD).

  6. The decreased RMSSD algorithm can be used to filter additional triggers that can be generated from other movisensXS blocks. That means you can e.g. have interspersed triggers when no low RMSSD trigger occurs. To send an additional trigger to the algorithm use a Send intent block with Action set to rmssd.random.
    Send Intent rmssd.random
    You can set the minimal time that is needed after a low RMSSD minute to accept an additional trigger. To determine which type of trigger was fired by the DecreasedRmssdAlgorithm, PRD will be shown for low RMSSD triggers and RND for interspersed triggers in the results file.

  7. The algorithm has two modes for triggering

    • direct mode: as soon as the algorithm detects a prolonged time in Low RMSSD as described above
    • externally triggered by mutable value: the algorithm keeps track of trigger events due to prolonged time in Low RMSSD in a history. The algorithm triggers only if it is externally triggered by a mutable value set to 1 and additionally having at least one trigger event in the history. History length can be defined.

# Parameter description

# Participant parameters

As previously noted, this algorithm requires a period of pre-study to obtain the individual RMSSD, Slope, and Intercept parameters of each participant. These parameters have to be entered after coupling a smartphone to the study:

Parameter Description
Regression Parameter slope in [ms] Slope used in the RMSSD prediction function
Regression Parameter intercept [ms/mg] or [mg/ms] Intercept used in the RMSSD prediction function
Threshold for RMSSD in [ms] Threshold used to determine if a minute is considered as low RMSSD

# Study parameters

The following parameters can be set in the DereasedRmssd block in the sampling scheme:

Parameter Default value Description
Regression method linear Select linear or inversion prediciton function
Window length of Movement Acceleration [min] 5min Window length to determine movement acceleration values for prediciton function (maximum)
RMSSD Assessment window in [min] 10min Window length to be used for decreased RMSSD assessment
No. of periods in Low RMSSD [min] 8min Number of periods in assessment window that have to be marked as decreased RMSSD to fire a trigger
Minimal distance between prompts in [min] 60min Minimal time between two triggers sent from this block regardless of low RMSSD trigger or additional interspersed trigger
Minimum time from last Low RMSSD [min] 10min If interspersed triggers are used, this parameter defines the minimal time for a interspersed trigger after a low RMSSD minute
Mutable Value for External Triggering empty If no mutable value is set, the algorithm runs in direct mode. If a mutable value is set, the algorithm runs in externally triggered by mutable value mode. The mutable value has to be set to 1 by a change mutable value block. Then the events in the history will be evaluated and the mutable value will be reset to 0.
History Length 20min Defines the length (in min) of the history which holds trigger events

# Literature

Last Updated: 6/10/2022, 1:25:46 PM
© 2024 movisens GmbH, Imprint