SYSTEMS AND METHODS FOR MEASURING ENERGY EXPENDITURE
Various methods and apparatuses for measuring a state parameter of an individual using signals based on one or more sensors are disclosed. In one embodiment, a first set of signals is used in a first function to determine how a second set of signals is used in one or more second functions to predict the state parameter. In another embodiment, first and second functions are used where the state parameter or an indicator of the state parameter may be obtained from a relationship between the first function and the second function. The state parameter may, for example, include calories consumed or calories burned by the individual. Various methods for making such apparatuses are also disclosed.
1 . (canceled)
2 . A method of making software for an apparatus for measuring energy expenditure of an individual, comprising:
providing a first sensor device, said first sensor device receiving a plurality of signals from at least two of a body motion sensor, a heat flux sensor, a skin conductance sensor, and a skin temperature sensor;
using said first sensor device to create one or more functions that predict said energy expenditure of said individual using a first set of signals based on one or more of said plurality of sensor signals; and
creating said software including instructions for:
(i) receiving a second plurality of signals collected by a second sensor device substantially structurally identical to said first sensor device for a period of time, said second sensor device receiving said second plurality of signals from at least two of a body motion sensor, a heat flux sensor, a skin conductance sensor, and a skin temperature sensor; and
(ii) utilizing a second set of signals based on one or more of said second plurality of sensor signals in said one or more functions to predict said energy expenditure of said individual.
3 . A method according to claim 2 , said apparatus comprising said second sensor device, said method further comprising storing said software in said second sensor device, said second sensor device having a processor for executing said software.
4 . A method according to claim 2 , said apparatus comprising said second sensor device and a computing device in electronic communication with said second sensor device for receiving said second plurality of signals from said second sensor device, said method further comprising storing said software in a computer readable medium for subsequent transfer to said computing device, said computing device having a processor for executing said software.
5 . A method according to claim 2 , wherein said step of using said sensor device to create said one or more functions includes gathering a first set of said plurality of signals under conditions where energy expenditure data for said individual is present, contemporaneously gathering gold standard data relating to said energy expenditure data for said individual, and using one or more machine learning techniques to generate said one or more functions from said first set of said plurality of signals and said gold standard data.
6 . A method according to claim 2 , said first sensor device receiving said plurality of signals from a body motion sensor, a heat flux sensor, and a skin conductance sensor.
7 . A method according to claim 6 , said body motion sensor being an accelerometer and said skin conductance sensor being a GSR sensor.
8 . A method according to claim 2 , said utilizing instruction comprising utilizing said second set of signals in a first function, said first function determining how a third set of signals based on one or more of said second plurality of sensor signals is utilized in one or more second functions, each of said one or more second functions having an output; wherein one or more of said outputs are used to predict said energy expenditure of said individual.
9 . A method according to claim 8 , wherein said first function recognizes one or more contexts based on said second set of signals, wherein one or more of said second functions is chosen based on said one or more recognized contexts, and wherein said outputs of said chosen second functions are used to predict said energy expenditure of said individual.
10 . A method according to claim 8 , wherein said first function recognizes each of a plurality of contexts based on said second set of signals, wherein each of said one or more second functions corresponds to one of said contexts, wherein said first function assigns a weight to each of said one or more second functions based on a recognition probability associated with the corresponding context, and wherein said outputs of said one or more second functions and said weights are used to predict said energy expenditure of said individual.
11 . A method according to claim 8 , said utilizing instruction further comprising combining said outputs in a post processing step to predict said energy expenditure of said individual.
12 . A method according to claim 8 , wherein said second functions are regression algorithms.
13 . A method according to claim 10 , wherein said contexts comprise rest and active.
14 . A method according to claim 13 , said first function comprising a naive Bayesian classifier.
15 . A method according to claim 13 , said receiving instruction comprising receiving said second plurality of sensor signals from a body motion sensor, a heat flux sensor, and a skin conductance sensor, said third set of signals comprising a heat flux high gain average variance (HFvar), a vector sum of transverse and longitudinal accelerometer SADs (YSAD), and a galvanic skin response low gain (GSR), wherein said second functions have the form of A*YSAD+B*HF+C*GSR+D*BMR+E, wherein A, B, C, D and E are constants and BMR is a basal metabolic rate for said individual.