An alarm algorithm was developed to monitor the ventilator on the National Aeronautics and Space Administration space station. The algorithm automatically identifies and interprets critical events so that an untrained user can manage the mechanical ventilation of a critically injured crew member. The algorithm was tested in two healthy volunteers by simulating 260 critical events in each volunteer while the volunteer breathed via the ventilator. Thirteen critical events were induced eight times in random order, for the five different modes of ventilation. These events included various ventilator tubing disconnects, leaks, and occlusions, as well as power and gas supply failures. The algorithm identified the critical events and generated alarms in response to 99.2% (516 of 520, total) of the events. The alarm textual messages were correct 98% (505 of 516 messages) of the time. The alarm algorithm is an improvement over current alarms found on most ventilators because its alarm messages specifically identify failures in the patient breathing circuit or ventilator. The system may improve patient care by helping critical care personnel respond more rapidly and correctly to critical events.

Algorithms↗

PubMed1989

The Utah anesthesia workstation.

A prototype anesthesia workstation has been developed to demonstrate the feasibility of a computer-assisted anesthesia workplace. The workstation provides a central display of information and aids the user in controlling and monitoring the anesthesia delivery system. The anesthesiologist interacts with the workstation through a Macintosh computer, which is easy for the clinician to understand and to use. Seventeen sensors and monitors transmit information from the anesthesia delivery system to the computer. The computer monitors this information using a set of rules, evaluated once each breath, to detect changes in the delivery system. If an event is detected, the computer alerts the anesthesiologist with a diagram, a text message, and an audible warning. A laboratory test of the monitoring system was performed to see if it properly identified 26 different critical events during simulated low flow and closed circuit anesthesia. Five hundred and eight-three of 660 simulated critical events (88%) were identified with the unique and correct message. On 35 occasions, multiple messages were displayed, including the correct one. Critical events were misidentified or not detected 42 times. Eight false positive alarms occurred during the 20 h of testing; all occurred as a result of baseline drift in a single transducer. These results demonstrate that a sophisticated monitoring system can reliably diagnose specific anesthesia machine failures.

Anesthesiology↗

PubMed1988

How the rise time of carbon dioxide analysers influences the accuracy of carbon dioxide measurements.

Carbon dioxide measurements are not accurate, especially in children, if the response time of the carbon dioxide analyser is too slow and its output fails to reach the actual carbon dioxide concentration at the end of the breath. The distortion of the carbon dioxide waveform is a function of the "rise time" of the analyser. We have simulated an expired carbon dioxide curve and calculated the rise time required to measure accurately end-tidal carbon dioxide and VCO2 in adults and children. A rise time of 80 ms (10-70%) is sufficient to measure end-tidal carbon dioxide concentration with 5% accuracy in patients with rates of ventilation less than 100 b.p.m. and I:E ratios less than 2:1. We have measured the rise time of 11 commercially available carbon dioxide analysers and found that only six of the 11 responded quickly enough to be accurate for rates up to 100 b.p.m. All 11 responded rapidly enough to measure end-tidal carbon dioxide concentration with 5% accuracy when ventilatory rates were less than 30 b.p.m. To measure VCO2 with 5% accuracy, an analyser should have a rise time of 20 ms. Only one analyser met this specification. An analyser's rise time can be estimated clinically to within 10 (SD 8) ms by a simple breath hold and forced exhalation, thus providing an estimate of the accuracy of carbon dioxide measurements in adults or children.

Anesthesia, Inhalation↗

PubMed1986

Gas exchange during mechanical ventilation and spontaneous breathing. Intermittent mandatory ventilation after open heart surgery.

Pulmonary gas exchange rates in eight patients after open heart surgery were studied during weaning from the ventilator. We investigated continuous positive pressure ventilation (CPPV), intermittent mandatory ventilation (IMV) and spontaneous breathing with continuous positive airway pressure (CPAP). During each mode of ventilation we measured: CO2 production (VCO2), O2 consumption (VO2), cardiac output (CO), PaO2, Qs/QT and functional residual capacity (FRC). In addition, we analyzed in each single breath: tidal volume (VT), series dead space volume (Vds), alveolar ventilation, alveolar efficiency for CO2 elimination (alv eff CO2) and end-tidal CO2 concentration (FCO2et). We compared the results of CPPV, IMV and CPAP and the mandatory breaths (MB) with the spontaneous breaths (SB) measured during IMV. CO was low during CPPV, when the patient still deeply sedated; it was increased in IMV and remained constant in the following CPAP period. VCO2 and VO2 did not differ significantly when switching from IMV to CPAP; therefore, work due to breathing seemed not to be reduced by the mandatory breath during IMV. Oxygenation (PaO2, Qs/QT) did not change significantly when switching from one mode to the other. FRC was constant when changing from CPPV to IMV, did not alter within the IMV-cycle and was reduced significantly when switching from IMV to CPAP. Dead space ventilation was reduced in SB (compared to MB). The latter result is discussed on the basis of two mechanisms: Vds was reduced and alv eff CO2 was increased. We conclude that compared to CPPV, IMV decreases mean alveolar pressure and reduces dead space ventilation at constant FRC and with constant oxygenation. This may explain why, in the weaning process, IMV makes it possible to start spontaneous breathing very early.

Adult↗

PubMed1985

A simple method for estimating compliance.

In intensive care medicine, pulmonary compliance is one of the very helpful diagnostic indices. Because of technical difficulties, however, the measurement of pulmonary compliance is often reduced to a rough guess of the compliance of the total respiratory system. The technical problems can be overcome using a computer to solve the basic equations with the least-squares fit (LSF) method. Unfortunately, this method requires such a long calculation time that bedside breath-by-breath calculations are impracticable on small computers. A simple computer algorithm (mean-values method) was therefore developed and compared to the LSF method. Compliance values calculated by either procedure were practically identical in ventilated patients. However, by reducing computing time to 30% of the LSF method, our mean-values algorithm enabled real-time estimation of compliance breath-by-breath.

Humans↗

PubMed1985

Accurate measurement of N2 volumes during N2 washout requires dynamic adjustment of delay time.

Measurement of respiratory gas composition by a mass spectrometer lags behind the measurement of gas flow. To obtain specific gas volumes (e.g., the N2 volume) by multiplication and integration of concentration and flow, one has to synchronize flow and concentration signals using the delay time (TD) of the gas analyzer. During the N2 washout, however, gas composition changes and causes alterations of TD. This leads to errors of up to 17 and 70% in the measurement of pulmonary volume and series dead space, respectively, in an ideally mixing physical model of the lung. On the basis of Poiseuille's law and exact measurements of the characteristics of the capillary it is possible to adjust the synchronization, which improves the absolute accuracy considerably.

Animals↗

PubMed1984

Series dead space volume assessed as the mean value of a distribution function.

Series dead space (VdS) is assumed to be represented by that volume exhaled until alveolar gas is observed. Phase II of the single breath CO2-diagram contains the (flow, concentration and sequence weighted) distribution off all stationary interfaces (SI) expired before phase III. We describe a new method to estimate the mean value of VdS based on the differentiation of phase II. This approximation of VdS is called the 'Pre Interface Expirate' (PIE) and is compared in this study with the integrative approach of Langley. Tidal volume (Vt) and PEEP were varied from 71 to 123% and from 0 to 6 cmH2O respectively. The estimation of VdS by differentiation of phase II (PIE) shows excellent reproducibility and depends only on phase II - not on phase III and IV as does VdS-Langley. PIE does not depend on Vt and PEEP per se but reflects the distension of convective airways due to elevated end-inspiratory airway pressure. Our results confirm the predictions of Paiva's model calculations in that the size of VdS is determined by the distension of airways rather than by the altered position of the SI.

Computers↗

PubMedDate not supplied

Knowledge engineering using retrospective review of data: a useful technique or merely data dredging?

The process of extracting the knowledge or rules for medical decision making is not an easy task. One approach to knowledge engineering is to carefully review how decisions were made in the past with the goal of extracting the rules. The purpose of this project was to use previously collected data from ICU patients to derive the rules for the definition of hemodynamic stability. 97 ICU patients between 9/9/86 and 7/29/90 were included in the analysis. All of these patients had adult respiratory distress syndrome. Their mechanical ventilation was managed by a set of computerized protocols. We retrospectively searched the HELP system database for instructions that were not followed due to hemodynamic reasons. For each patient, we also chose one randomly selected therapy instruction which was followed to act as a control. For each instruction we then selected the corresponding hemodynamic data set. The data was then used in a stepwise logistic regression to determine the rules used for defining hemodynamic instability. We found that several of the hemodynamic parameters we had anticipated to be important were not even measured most of the time. The blood pressures and heart rate were almost identical between the hemodynamicly stable and unstable data sets. We conclude that the decision making process used by physicians has great variation, both between and within physicians. This makes knowledge engineering using retrospective techniques such as this prone to error and probably not very fruitful.

Bias↗