Measuring psychological stress from cardiovascular and activity signals
A method and system for measuring psychological stress disclosed. In a first aspect, the method comprises determining R-R intervals from an electrocardiogram (ECG) to calculate a standard deviation of the R-R intervals (SDNN) and determining a stress feature (SF) using the SDNN. In response to reaching a threshold, the method includes performing adaptation to update a probability mass function (PMF). The method includes determining a stress level (SL) using the SF and the updated PMF to continuously measure the psychological stress. In a second aspect, the system comprises a wireless sensor device coupled to a user via at least one electrode, wherein the wireless sensor device includes a processor and a memory device coupled to the processor, wherein the memory device stores an application which, when executed by the processor, causes the processor to carry out the steps of the method.
1. A method for adaptive function calibration, comprising:
initializing a stress feature probability density function (PDF) to a predetermined group model;
determining a resting heart rate (HR rest ) of a user during a period of low or no activity;
estimating an individual heart rate (HR) PDF;
estimating an individual standard deviation of an R-R interval (SDNN) PDF;
computing a new stress feature PDF by combining the determined HR rest and SDNN PDF; and
converting a continuous PDF SF to at least one of: a discretized probability distribution or a probability mass function (PMF SF ).
2. The method of claim 1 , wherein the initializing the stress feature PDF to the predetermined group model stress feature is by: PDF SF ˜N(μ SFgroup , σ 2 SFgroup ).
3. The method of claim 2 , wherein the notation N(μ,σ 2 ) is a normal/Gaussian distribution with mean μ and variance σ 2 .
4. The method of claim 3 , wherein the μ SFgroup and σ 2 SFgroup are predetermined from group training data.
5. The method of claim 1 , wherein the HR rest is estimated from user data.
6. The method of claim 5 , wherein the low or no activity includes sleep periods.
7. The method of claim 1 , wherein the individual HR PDF is determined by: PDF HR ˜N(HR rest +γ*σ HRindiv , σ 2 HRindiv ).
8. The method of claim 1 , wherein the individual SDNN PDF is estimated by: PDF SDNN ˜N(μ SDNNindiv , σ 2 SDNNindiv ).
9. The method of claim 1 , wherein the new stress feature PDF is computed by combining the determined HR and SDNN PDFs by: PDF SF ˜N(HR rest +γ*σ HRindiv +α*μ SDNNindiv , σ 2 HRindiv +α 2 *σ 2 SDNNindiv ).
10. The method of claim 1 , wherein the converting the continuous PDFSF is done by sampling the PDF SF within a predetermined interval and normalizing a sum to 1.
11. A non-transitory computer-readable medium storing executable instructions that, in response to execution, cause a computer to perform operations comprising:
initializing a stress feature probability density function (PDF) to a predetermined group model;
determining a resting heart rate (HR rest ) of a user during a period of low or no activity;
estimating an individual heart rate (HR) PDF;
estimating an individual standard deviation of an R-R interval (SDNN) PDF;
computing a new stress feature PDF by combining the determined HR rest and SDNN PDF; and
converting a continuous PDF SF to at least one of: a discretized probability distribution or a probability mass function (PMF s1 ).