Patent Application
20070295713
A thermometer for measuring core body temperature includes a plurality of sensors and thermal insulation in a layered configuration. The thermometer is applied externally to the body being measured. A first of the sensors measures temperature at the skin and at least one other sensor measures temperature away from the skin. The temperatures measured by the sensors provide data for a calculation of temperature internal to the body.
In operation, a first sensor 105 of the plurality of sensors, is placed in contact with the surface of the skin 155 of the body 160 to be measured. The thermometer 100 is made up of a “sandwich” or “telescope” of temperature sensors 105, 115, 125, 135 and layers of insulative material 110, 120, 130, 140 where the insulative material 110, 120, 130, 140 provides both thermal isolation of the sensors from the ambient environment 105, 115, 125, 135 and thermal thickness for thermal attenuation. The core temperature of the body 160 can be determined through extrapolation given temperature measurements from the sensors 105, 115, 125, 135.
Four layers of sensors and insulation components are shown in
There are four elements that enable the external thermometer to operate in a core body temperature application. The thermometer array of the present invention encapsulates one or more temperature sensors so that they are embedded within a “sandwich” or “telescope.” This configuration has properties that enable the construction and use of an algorithm to extrapolate the core body temperature from the sensor temperature measurements in specified locations of the telescope.
A first element is the establishment of a strong coupling between the temperature sensor 105 and the skin 155 of the body 160 and a weak coupling of that same sensor 105 with the environment around the body 160. This is accomplished by the thermal insulation component 110 that surrounds the sensor 105 on all sides except the side of the sensor 105 that contacts the skin 155. The insulation component 110 has sufficient thickness at the sides of the sensor 105 so that there is substantially a negligible thermal contact with the environment around the thermometer 100. The edge areas are approximately 0.25 to 0.75 inches thick. The material composing the insulation component 110 has, in a first embodiment, a low thermal resistance between the thermal sensors, for example, an R-value equal to 1. This corresponding U-value is selected in order to provide thermal isolation from the environment even as the insulation component is used as a medium through which the temperature difference between the body and the environment can be attenuated. Examples of such materials are expanded semi-rigid rubber, or a neoprene blend with a nylon cover such as that used in knee supports for athletes. This material is impermeable to sweat and water so that there is evaporative cooling on the outside surface of the insulation but an absence of evaporative cooling from the immediate surface of the sensor array or the skin immediately underneath and in the immediate vicinity of the sensor array. In a preferred embodiment, the layers of the telescope are modularized with each sensor sitting in a small pocket of insulation that surrounds it on all but one side, this side being exposed to the previous layer and the layers fused together.
The protective layer 150 generally has a negligible insulative value. In one embodiment the protective layer 150 is not used. In another embodiment, the protective layer 150 would be present purely as a protective layer, separating the skin 155 from the first sensor 105.
The second element that enables the external thermometer is that of a thermal “sandwich” or “telescope” configuration. This is accomplished by successive layers of a temperature sensor encased in an insulative layer, then another temperature sensor encased in another insulative layer and so on. The layers can be iterated as many times as may be required to obtain more data points to make the application of the algorithm more accurate.
The third element is the blocking of the evaporation of sweat at the site on the skin 155 where the thermometer 100 is applied. Impermeability to moisture from the skin is accomplished by making the insulative thermometer housing 100 impermeable to sweat. This substantially eliminates the confounding effect of evaporation on the measurement of temperature in the region of the thermometer, by substantially eliminating the presence of sweat on the array surface, by making its effect predictable and measurable through calibration, or in ambient regimes conducive to profuse sweating, by the natural near-elimination of evaporative cooling on the array surface.
The fourth element is that of calibrating the thermometer. One example calibration point is an extreme case where a single thermal sensor is placed on the skin surface and surrounded by R-30 insulation, so that the single thermal sensor couples strongly to the body heat reservoir and very weakly to the ambient air reservoir. In this configuration, the thermometer measures temperatures that are effectively internal temperatures of the body. The surface skin has thickness including fatty layers and other characteristics that typically vary from one person to another. Thus the effective R-value of the skin typically varies from one person to another. The effective R-value of this skin layer can be calculated and factored into an algorithm or graph by measuring the temperatures at the layers of the thermal sensor telescope as well as the core temperature of the individual. The graph is of a type similar to that shown in
The thermometer device of the present invention is generally particularly accurate in certain regimes of interest. One such regime is that of humans such as athletes, functioning in environmental temperatures near the body temperature of 98.6° F. In contrast, the temperature difference between an exposed body part and arctic environments is likely to be large, leading to the need for large insulative layers to couple the temperature sensors to the body rather than to the environment. Such large insulative layers can include a mountaineering parka, with the thermometric device entirely on the body side of the parka and measuring core temperature. In an environment having an ambient temperature close to normal body temperature, the temperature difference is small, the slope of the graph of temperature versus position is essentially flat, leading to the requirement of a minimum amount of insulation to isolate the thermometric device. Thus, if the purpose of the thermometer device is to measure whether an active person is approaching a dangerously high core temperature, such as 101.5° F., at which one is at risk for the onset of heat exhaustion, the accuracy of the external core temperature measurement is readily sufficient, even with a very small device with minimal insulation.
The thermometer 100 is held close to the skin 155 by an adhesive in a first embodiment. In a second embodiment, the thermometer 100 is held close to the skin 155 by embedding the thermometer 100 in a belt or strap, such as a chest strap or a shoulder strap as shown in
In further alternative embodiments, the thermometer 100 is embedded subcutaneously. In the subcutaneous embodiment, the details of the temperature dynamics and the algorithm are appropriately altered to account for the differences in environment in this configuration. In this embodiment, the skin surface itself contains the thermometer as described herein, the tissue of the skin provides insulation and isolation, and the calibration and analysis properties of the thermometer are suitably modified.
A telescope of insulative layers in the thermometer is analogous to the layers of thermal insulation on a building. The successive insulative layers of a house may consist, for example, of clapboard, tar paper, sheathing, two-by-fours with glass fiber insulation in the air spaces, then vapor barrier and wallboarding. Each of the layers provides a separate insulative layer, attenuating the temperature difference between indoors and outdoors respectively. If the indoor temperature is, for example, 75° F. and the outdoor temperature is 35° F., then the temperature measured at any point in the wall is somewhere between 75° F. and 35° F. The temperature measurement varies as one moves inward or outward through the materials. If there is adequate insulation, the temperature throughout most of the structure is independent of whether there is some surface evaporative cooling on the outside wall. Further, the temperature varies approximately linearly as a proportion of the resistance value of the materials from the wall to a given point to the total insulative resistance value multiplied by the temperature difference and added to the beginning temperature. Knowledge of the temperature outdoors along with the shape of the resistance curve allows one to extrapolate the indoor temperature as a function of temperature measurements made outside the interior. If the wall were extended in certain locations consistent with the guidelines of the above teaching, and the thermometric array calibrated consistent with the insulation value of the wall, the same physics would still hold true, and it would be possible to extrapolate the indoor temperature of the house from information gathered entirely outside the house.
T
n=(Tcnv−Tcore)*(Rn/Rtot) (1)
where
Rn=R
2
+ . . . +R
n-1, and
Rtot=R
2
+ . . . +R
6.
The experimental subject in the experiments generating the data shown in
For greater/lesser increase in the internal rate of heat generation, the instantaneous temperature difference between the innermost layers and the outer layers is greater/lesser, so that diffusion is greater and slightly more/less time is required to reach equilibrium. For lesser insulation values for each of the layers, there is a decrease in the lag time before equilibrium, when the most accurate core temperature can be determined using this method. Prior to equilibrium, it is possible to measure core temperature externally or internally, but this measurement is not as accurate as when the measurement is performed after equilibrium is attained. Typically, a best time to measure rapid increases in core body temperature is prior to equilibrium, because the first derivative of temperature with respect to time is a measure of the differential between core temperature and the known initial temperature at the measurement point.
Returning to
For each activity regime and set of layers for the temperature measurement, there is a temperature measurement. In each case, the measured temperature rises along a smooth curve as predicted from the temperature at the outermost layer (which is closer to the core body temperature than the reading of the environmental temperature because there is a layer of insulation between the environment and the first temperature sensor so that the sensor is coupled somewhat more to the body temperature and somewhat less to the ambient temperature) to the core body temperature. As the ambient temperature through the various layers increases toward the core body temperature, the curve flattens. When the environmental temperature and the core body temperature are the same, the slope of the curves is zero. Obviously then, it is for environmental temperatures in the upper 90s ° F. and lower 100s ° F. that the method most readily provides accurate temperatures for either active or inactive people. When the environmental temperature is greater than the core body temperature, the slope of the curve becomes positive, indicating that the thermal layers are insulating the body against the heat. Again it is worth emphasizing that each of these curves is continuous, analytic and can be extrapolated to fill in any single missing data point, no matter where that point may be in the geometry of the configuration. Further the set of curves provides a continuously varying set for which any one member can be interpolated in relation to other curves immediately above it or below it on the graph.
Equation (1) predicts the temperature curves, and conversely, either equation (1) or the temperature curves can be used as the means to extrapolate the core body temperature given the temperatures at TR1 and TR2.
Statistical learning techniques can also be used to “phase lock” the incoming data stream and to interpret the pattern of temperatures at the various layers of the sandwich. Statistical learning techniques are particularly useful for interpreting the pre-equilibrium data from the temperature sensors in the inventive device.
The datastream from the individual thermometers includes time varying values received by the controller and classifier engine as a result of various factors: including system noise, small variances in the thermometers and electronic components, temporary malfunction, physical or other shocks to the system. In one embodiment of the system the datastream values are fed through a classifier in order to “phase lock” the system so that such momentary variances are smoothed out to produce a continuous curve. A classifier is a predictive model that predicts an output value based on input data appropriate to the model. Throughout this discussion the general term “model” or “model/classifier” is used herein to describe any type of signal processing or analysis, statistical modeling, regression, classification technique, or other form of automated real-time signal interpretation. If a variance persists as indicative of such as equipment failure, then the curve deflects from its otherwise normal path and a readily recognized alert is switched on. In another embodiment, the classifier(s) used in determining a core temperature from the array temperatures are trained on noisy data so that variations in the input datastream are a normal part of the data analysis and do not confuse or lead to an abortion of the process.
Because the layered sensors enable the extrapolation of core body temperature and also provide differentials with respect to space and time, statistical learning techniques can be applied to rich data. For example, a Bayesian classifier can learn patterns of interest in the temperatures and differentials, and predict likely future troubles with core body temperature, while at the same time providing a probability that these troubles will occur. The following example is provided to illustrate the present invention using Bayesian classifiers. Other types of classifier are considered to be within the scope of the invention. The present invention is applicable in other situations.
In one case, that of linemen of professional football teams, there is iterated extreme exertion for five to fifteen seconds followed by a rest period for twenty to thirty seconds. Thus the core body temperature sensor will seldom be in equilibrium during the course of a practice or game. The pattern of temperature values, however, at the various layers can be learned by the classifier. There will be temperature sequences that will be benign, and other temperature sequences that will signal likely overheating. These can be captured using sample training data for insertion into a model for a statistical classifier. As one example of such a machine learning scenario, if the lineman's core body temperature is 100.5° F. and the lineman exerts himself more excessively during a hurry-up offense with shorter breaks, on the order of 0-10 seconds, then the temperatures in the inner layers will increase more rapidly than those in the outer layers, and also will increase differentially in comparison to the increase when there is less exertion or a longer rest period between exertions. Sequences such as this that lead to troubles, and sequences that do not lead to troubles, can be identified experimentally by instrumenting lineman while they perform their normal, or artificially constructed, activities. A Bayesian classifier, for example, can learn from individual linemen's profiles, or from learning samples of linemen's profiles, the patterns for the individual or a class of linemen that will lead to high risk of heat exhaustion versus little risk of heat exhaustion. While the learning time for creating a model for such classifiers can be substantial, the operation time to apply the model can be quite short. Thus, in substantially real time, the coaches and trainers on the sideline can know whether a particular player is at significantly increasing risk of heat exhaustion (and also resulting performance decrement) before his core temperature reaches 101.5° F. Knowing in advance can enable substitution patterns during the play of the game that reduces risk and gives a competitive advantage. Knowing in advance can also lead to simple ameliorative actions during convenient times (such as a player standing in front of a fan to cool off more for some downs when already off the field, and then returning to the game or practice) rather than the more serious actions that may be required (such as missing several series of downs, or the remainder of the game) when the player begins to exhibit symptoms of heat exhaustion.
In a first embodiment of the temperature measuring system 300, the analytics device 305 and the output device 330 are integral to the thermometer 100. The output device 330 is for example a simple readout device such as an liquid crystal display (LCD). Alternatively, the output device 330 is a transmitter that transmits data to an external receiving device. Typically, the external receiving device has an associated display. In a further alternative embodiment, the analytics device 305 is not integral to the thermometer 100. In this embodiment, the thermometer 100 incorporates a transmitter capable of transmitting sensor data to the analytics device 305. In a still further alternative embodiment, the analytics device 305 communicates with a data collection device 335, which is typically not integral to the thermometer 100, and may be on the sideline or accessed through the Internet or by other means; however an integral data collection device or a data collection device 335 worn on the body of the person wearing the thermometer 100 is possible.
At step 410, the analytics device analyzes the plurality of temperature measurements. In a first embodiment, the analytics device applies an extrapolation algorithm in order to obtain temperature values such as the core body temperature from the measured temperatures. In an alternative embodiment, the analytics device graphs the temperatures. In another alternative embodiment, the analytics device performs a statistical classification recognition as described above.
At step 415, the analytics device produces a derived temperature value based on the results of step 410. The analytics device at this step produces a derived temperature for a layer other than the layers measured in step 405, for example, a core body temperature. In a first embodiment, the derived core body temperature is a result of extrapolation from the temperature measurements taken in step 405 and analyzed in step 410. In a second embodiment, the derived core body temperature is a result of a prediction made from graphed data. In a third embodiment, the core body temperature is produced from a model determined through statistical analysis applied in step 410.
Using the analytics device as described above, data from the layered thermometer according to embodiments of the invention can be used to determine temperature at a layer outside of the layered thermometer such as core body temperature. As described above, in a first arrangement, the analytics device is integral to the thermometer. In another arrangement, the thermometer and the analytics device are separate devices that communicate wirelessly for example although a wired connection is possible for some applications.
It is to be understood that the above-identified embodiments are simply illustrative of the principles of the invention. Various and other modifications and changes may be made by those skilled in the art which will embody the principles of the invention and fall within the spirit and scope thereof.
