Semicontinuous cardiac output monitoring using a neural network

Objectives: This study compared 2-mL bolus thermodilution cardiac output me asurements with standard 10-mL bolus measurements, Design: Cardiac output was measured with the new 2-mL bolus technique and t he 10-mL standard thermodilution technique in a perspective series. We desc ribe a system that automatically cools and injects 2-mL boluses of saline i nto a standard pulmonary artery catheter, It uses a Peltier effect solid-st ate cooler and pneumatically driven syringe injector to measure cardiac out put once per minute, Setting: Animal laboratory, Animals: Eight adult Duroc swine weighing between 38.0 and 57.5 kg, Interventions., Once each minute, 2 mt of cooled 5% dextrose was injected t hrough the pulmonary catheter, Once every 8 mins, four sequential measureme nts of cardiac output were made using 10-mL injections, Measurements and Main Results: A total of 1249 paired waveforms were proces sed with both a conventional algorithm and with a neural network, For the c onventional algorithm, the correlation coefficient was r(2) = .92 and the s o of the difference was 1.30 L/min. For the neural network, the correlation coefficient was r(2) = .94 and the so of the difference was 0.88 L/min, Ou tput filtering improved the results in both cases. Conclusion: Neural networks accurately derive cardiac output from 2-mL bolu s thermodilution injections, allowing cardiac output to be monitored automa tically once per minute in many patients, The technique is convenient and u ses standard low-cost catheters.