Quantifying the accuracy of physiological data measured by a Vital Signs Detection System (VSDS) plays a key role in
making trustworthy decisions about the physiological status of a soldier. We developed an algorithm to report VSDSmeasured
heart and respiratory rates and their associated confidence levels. Heart and respiratory rates were measured
about every 2 seconds for about 4 hours, while subjects engaged in low (e.g., sitting), medium (e.g., sit-ups), and high
intensity (e.g., running) activities. The mean heart and median respiratory rates are calculated every 15 seconds by an
in-house developed algorithm, and associated confidence levels for each variable are estimated simultaneously using a
fuzzy-logic-based algorithm. Inputs into the algorithm are features that represent two types of information; the quality
of each variable, and the relationship between the variables. Faulty data points are separated from good measures by
setting a threshold. When data with pre-classified faults are tested with the confidence level threshold set at 0.5, the
sensitivity and specificity of the algorithm for heart rate are 91% and 97%, respectively. For respiratory rate, because of
the intrinsically noisy property of the data, the sensitivity and specificity are 87% and 93%, respectively. These
preliminary results demonstrate that the fuzzy logic algorithm can accurately qualify heart and respiratory rates
measured by a VSDS.
Respiratory waveforms and their derived respiratory rate time-series data can become misaligned from each other when they are collected by vital signs monitors under sub-optimal field conditions. The monitor-provided waveforms and rates can be re-aligned by independently calculating respiratory rates from the waveforms and then aligning them with the monitor-provided rates. However, substantially different rates may be generated from the same waveform due to the presence of ambiguous breaths at noisy positions in the waveform. This paper reports a landscape matching (LAM) algorithm to align respiratory rate time-series data with the waveform that they are derived from by using rates that are calculated by different means. The algorithm exploits the intermittent matches between two respiratory rate time series to generate a matching score for an alignment. The best alignment exhibits the highest matching score. The alignment performance of the LAM algorithm is compared to that of a correlation matching (CM) algorithm using field-collected respiratory data. Alignment performance is evaluated by: (1) comparing the ability of the two algorithms to return a shifted waveform to its original, known position; and (2) comparing the percent of points that match between the monitor-provided and calculated respiratory rate time-series data after re-alignment. The LAM alignment algorithm outperforms the CM algorithm in both comparisons at a statistically significant level (p<0.05). Out of 67 samples with shifted time-series data, on average, the LAM aligns respiratory rates within 44 seconds of the original position, which is significantly better the CM-calculated alignment (136 seconds). Out of 465 samples, the LAM performs better, worse, and equal to the CM algorithm in percentage of points matching in 73%, 11%, and 16% of the cases, respectively. This robust alignment algorithm supports the use of reliable post-hoc monitor-provided respiratory rates for data mining purposes.
This paper explores how the accuracy of a first-principles physiological model can be enhanced by integrating data-driven, "black-box" models with the original model to form a "hybrid" model system. Both linear (autoregressive) and nonlinear (neural network) data-driven techniques are separately combined with a first-principles model to predict human body core temperature. Rectal core temperature data from nine volunteers, subject to four
30/10-minute cycles of moderate exercise/rest regimen in both CONTROL and HUMID environmental conditions, are used to develop and test the approach. The results show significant improvements in prediction accuracy, with average improvements of up to 30% for prediction horizons of 20 minutes. The models developed from one subject's data are also used in the prediction of another subject's core temperature. Initial results for this approach for a 20-minute horizon show no significant improvement over the first-principles model by itself.