Improving Real-Life Estimates of Emotion Based on Heart Rate: A Perspective on Taking Metabolic Heart Rate Into Account

Figures

• 

  Figure 1. Overview of how the most important studies discussed in this article are embedded among the different variables used to estimate (non-metabolic) heart rate. Numbers preceding the names of the studies’ first authors refer to the sections in which they are discussed. In addition, the proposed next step is depicted.

• 

  Figure 2. Example data of overall dynamic body acceleration (ODBA) and heart rate over 24 h in an animal. Interruptions in the data are caused by interruptions in the heart rate recording. (Figure 3 in Miwa et al., 2015; reprinted with permission).

• 

  Figure 3. Example data of body movement, heart rate and derived mental arousal level (MAL) of a participant walking home from work. The data was interpreted as the participant being a little excited to leave work in the beginning and being irritated at second 1250 because of a phone call. The continued positive MAL after the participant arrived at home and sat on the sofa (second 1360) was interpreted as the participant thinking about the phone call (Figures 1, 4 in Yang et al., 2017; reprinted with permission).

• 

  Figure 4. Raw heart rate, baselined heart rate, additional (non-metabolic) heart rate and subjective emotion during respectively pre-flight baseline as recorded in a hospital; walking from the hospital to the car; walking from the terminal to the airplane; flight; landing; post-flight baseline. Data from flight phobic participants are represented by squares and a solid line; data from control participants are represented by circles and dashed lines (Modified Figure 2 in Wilhelm and Roth, 1998b; reprinted with permission).

• 

  Figure 5. Root-mean-square-error (RMSE) of energy expenditure estimation following the traditional “counts based” method for sedentary and active clusters (the two data distributions on the right, marked CS and CA); and the “activity specific” method as a function of the number of used accelerometers, for both sedentary (dark gray) and active (light gray) behavior (Figure 4B in Altini et al., 2015; reprinted with permission).