John Mark Ansermino

Multiparameter Respiratory Rate Estimation From the Photoplethysmogram

We present a novel method for estimating respiratory rate in real time from the photoplethysmogram (PPG) obtained from pulse oximetry. Three respiratory-induced variations (frequency, intensity, and amplitude) are extracted from the PPG using the Incremental-Merge Segmentation algorithm. Frequency content of each respiratory-induced variation is analyzed using fast Fourier transforms. The proposed Smart Fusion method then combines the results of the three respiratory-induced variations using a transparent mean calculation. It automatically eliminates estimations considered to be unreliable because of detected presence of artifacts in the PPG or disagreement between the different individual respiratory rate estimations. The algorithm has been tested on data obtained from 29 children and 13 adults. Results show that it is important to combine the three respiratory-induced variations for robust estimation of respiratory rate. The Smart Fusion showed trends of improved estimation (mean root mean square error 3.0 breaths/min) compared to the individual estimation methods (5.8, 6.2, and 3.9 breaths/min). The Smart Fusion algorithm is being implemented in a mobile phone pulse oximeter device to facilitate the diagnosis of severe childhood pneumonia in remote areas.

Current CHS and NHBPEP Criteria for Severe Preeclampsia Do Not Uniformly Predict Adverse Maternal or Perinatal Outcomes

Objective: To determine the association between adverse maternal/perinatal outcomes and Canadian and U.S. preeclampsia severity criteria. Methods: Using PIERS data (Preeclampsia Integrated Estimate of RiSk), an international continuous quality improvement project for women hospitalized with preeclampsia, we examined the association between preeclampsia severity criteria and adverse maternal and perinatal outcomes (univariable analysis, Fishers exact test). Not evaluated were variables performed in <80% of pregnancies (e.g., 24-hour proteinuria). Results: Few of the evaluated variables were associated with adverse maternal (chest pain/dyspnea, thrombocytopenia, ‘elevated liver enzymes’, HELLP syndrome, and creatinine >110 μM) or perinatal outcomes (dBP >110 mm Hg and suspected abruption) (at p < 0.01). Conclusions: In the PIERS cohort, most factors used in the Canadian or American classifications of severe preeclampsia do not predict adverse maternal and/or perinatal outcomes. Future classification systems should take this into account.

Photoplethysmogram signal quality estimation using repeated Gaussian filters and cross-correlation

Pulse oximeters are monitors that noninvasively measure heart rate and blood oxygen saturation (SpO2). Unfortunately, pulse oximetry is prone to artifacts which negatively impact the accuracy of the measurement and can cause a significant number of false alarms. We have developed an algorithm to segment pulse oximetry signals into pulses and estimate the signal quality in real time. The algorithm iteratively calculates a signal quality index (SQI) ranging from 0 to 100. In the presence of artifacts and irregular signal morphology, the algorithm outputs a low SQI number. The pulse segmentation algorithm uses the derivative of the signal to find pulse slopes and an adaptive set of repeated Gaussian filters to select the correct slopes. Cross-correlation of consecutive pulse segments is used to estimate signal quality. Experimental results using two different benchmark data sets showed a good pulse detection rate with a sensitivity of 96.21% and a positive predictive value of 99.22%, which was equivalent to the available reference algorithm. The novel SQI algorithm was effective and produced significantly lower SQI values in the presence of artifacts compared to SQI values during clean signals. The SQI algorithm may help to guide untrained pulse oximeter users and also help in the design of advanced algorithms for generating smart alarms.

Design and Evaluation of a Low-Cost Smartphone Pulse Oximeter

Infectious diseases such as pneumonia take the lives of millions of children in low- and middle-income countries every year. Many of these deaths could be prevented with the availability of robust and low-cost diagnostic tools using integrated sensor technology. Pulse oximetry in particular, offers a unique non-invasive and specific test for an increase in the severity of many infectious diseases such as pneumonia. If pulse oximetry could be delivered on widely available mobile phones, it could become a compelling solution to global health challenges. Many lives could be saved if this technology was disseminated effectively in the affected regions of the world to rescue patients from the fatal consequences of these infectious diseases. We describe the implementation of such an oximeter that interfaces a conventional clinical oximeter finger sensor with a smartphone through the headset jack audio interface, and present a simulator-based systematic verification system to be used for automated validation of the sensor interface on different smartphones and media players. An excellent agreement was found between the simulator and the audio oximeter for both oxygen saturation and heart rate over a wide range of optical transmission levels on 4th and 5th generations of the iPod Touch™ and iPhone™ devices.

Identification by families of pediatric adverse events and near misses overlooked by health care providers

Background: Identifying adverse events and near misses is essential to improving safety in the health care system. Patients are capable of reliably identifying and reporting adverse events. The effect of a patient safety reporting system used by families of pediatric inpatients on reporting of adverse events by health care providers has not previously been investigated. Methods: Between Nov. 1, 2008, and Nov. 30, 2009, families of children discharged from a single ward of British Columbia’s Children’s Hospital were asked to respond to a questionnaire about adverse events and near misses during the hospital stay. Rates of reporting by health care providers for this period were compared with rates for the previous year. Family reports for specific incidents were matched with reports by health care providers to determine overlap. Results: A total of 544 familes responded to the questionnaire. The estimated absolute increase in reports by health care providers per 100 admissions was 0.5% (95% confidence interval −1.8% to 2.7%). A total of 321 events were identified in 201 of the 544 family reports. Of these, 153 (48%) were determined to represent legitimate patient safety concerns. Only 8 (2.5%) of the adverse events reported by families were also reported by health care providers. Interpretation: The introduction of a family-based system for reporting adverse events involving pediatric inpatients, administered at the time of discharge, did not change rates of reporting of adverse events and near misses by health care providers. Most reports submitted by families were not duplicated in the reporting system for health care providers, which suggests that families and staff members view safety-related events differently. However, almost half of the family reports represented legitimate patient safety concerns. Families appeared capable of providing valuable information for improving the safety of pediatric inpatients.

A wavelet approach to detecting electrocautery noise in the ECG

A software approach has been developed for detecting electrocautery noise in the electrocardiogram (ECG) using a wavelet decomposition of the signal. With this approach, a clinical monitoring expert system can be forewarned of potential artifacts in trend values derived from the ECG, allowing it to proceed with caution when making decisions based on these trends. In 15 operations spanning 38.5 h of ECG data, we achieved a false positive rate of 0.71% and a false negative rate of 0.33%. While existing hardware approaches detect the source of the noise without any ability to assess its impact on the measured ECG, our software approach detects the presence of noise in the signal itself. Furthermore, the software approach is cheaper and easier to implement in a clinical environment than existing hardware approaches

Usability testing of a prototype Phone Oximeter with healthcare providers in high‐ and low‐medical resource environments*

To increase the use of pulse oximetry by capitalise on the wide availability of mobile phones, we have designed, developed and evaluated a prototype pulse oximeter interfaced to a mobile phone. Usability of this Phone Oximeter was tested as part of a rapid prototyping process. Phase 1 of the study (20 subjects) was performed in Canada. Users performed 23 tasks, while thinking aloud. Time for completion of tasks and analysis of user response to a mobile phone usability questionnaire were used to evaluate usability. Five interface improvements were made to the prototype before evaluation in Phase 2 (15 subjects) in Uganda. The lack of previous pulse oximetry experience and mobile phone use increased median (IQR [range]) time taken to perform tasks from 219 (160–247 [118–274]) s in Phase 1 to 228 (151–501 [111–2661]) s in Phase 2. User feedback was positive and overall usability high (Phase 1 – 82%, Phase 2 – 78%).

A Direct Dynamic Dose-Response Model of Propofol for Individualized Anesthesia Care

In an effort to open up new opportunities in individualized anesthesia care, this paper presents a dynamic dose-response model of propofol that relates propofol dose (i.e., infusion rate) directly to a clinical effect. The proposed model consists of a first-order equilibration dynamics plus a nonlinear Hill equation model, each representing the transient distribution of propofol dose from the plasma to the effect site and the steady-state dose-effect relationship. Compared to traditional pharmacokinetic-pharmacodynamic (PKPD) models, the proposed model has structural parsimony and comparable predictive capability, making it more attractive than its PKPD counterpart for identifying an individualized dose-response model in real-time. The efficacy of the direct dynamic dose-response model over a traditional PKPD model was assessed using a mixed effects modeling analysis of the electroencephalogram (EEG)-based state entropty (SE) response to intravenous propofol administration in 34 pediatric subjects. An improvement in the mean-squared error and r2 value of individual prediction, as well as the Akaikes information criterion (AIC) was seen with the direct dynamic dose-response model.

Adaptive Change Detection in Heart Rate Trend Monitoring in Anesthetized Children

The proposed algorithm is designed to detect changes in the heart rate trend signal which fits the dynamic linear model description. Based on this model, the interpatient and intraoperative variations are handled by estimating the noise covariances via an adaptive Kalman filter. An exponentially weighted moving average predictor switches between two different forgetting coefficients to allow the historical data to have a varying influence in prediction. The cumulative sum testing of the residuals identifies the change points online. The algorithm was tested on a substantial volume of real clinical data. Comparison of the proposed algorithm with Triggs approach revealed that the algorithm performs more favorably with a shorter delay. The receiver operating characteristic curve analysis indicates that the algorithm outperformed the change detection by clinicians in real time

Closed-Loop Anesthetic Drug Concentration Estimation Using Clinical-Effect Feedback

This letter presents a novel closed-loop approach to anesthetic drug concentration estimation using clinical-effect measurement feedback. Compared with the open-loop prediction used in current target-controlled infusion systems, closed-loop estimation exploits the discrepancy between the measured and predicted clinical effects to make corrections to the drug-concentration estimate, achieving improved robustness against variability in the patient pharmacokinetics and pharmacodynamics. A robust estimator, which processes drug administration and clinical-effect measurements to estimate the plasma- and effect-site drug concentrations, is designed using -synthesis theory. Initial proof of principle of the closed-loop estimation is demonstrated using the Monte Carlo simulation of surgical procedures with a wide range of patient models. Closed-loop estimation results in statistically significant reductions in median percentage, median absolute percentage, and maximum absolute percentage drug-concentration errors compared to open-loop prediction.