Device Control
Provision of Decision Support to a Closed-Loop Mechanical Ventilator
Scenario Details: Bobby’s clinical condition gradually stabilizes with antibiotics and blood pressure support. As his capillary leak and pulmonary edema improves, Bobby’s ventilator settings become excessive and his blood gases become dangerously alkalotic. His ventilator needs to be weaned, but his respiratory therapist is busy with a new admission who is also preoccupying David’s time.
Fortunately, Bobby’s ventilator is a new SmartVent, a machine compatible with the hospital CDS infrastructure and capable of maintaining a patient within desired clinical parameters with little or no human intervention. Medical equipment attached to Bobby (a pulmonary function machine, his vital sign monitor, etc) stream real-time data into the central decision analysis server. The system then determines the correct ventilator changes required to keep Bobby within safe blood gas values and provides those changes back to the SmartVent. In advisory mode, the SmartVent displays those recommendations on the screen. In closed-loop mode, it automatically adjusts itself. David had changed the ventilator from advisory to self-adjusting mode just that morning.
For most of the night, while his team was pre-occupied, SmartVent successful weaned his settings. Around 4 am, however, a ventilator adjustment did not result in the expected changes in his blood gas - in fact, it had the opposite effect. Confronted with unexpected and unexplained patient “behavior”, the CDS infrastructure pages the respiratory therapist. It then pages David and the charge nurse 10 min later when the RT fails to log into the SmartVent console to acknowledge the alert. When David arrives, he quickly determines that Bobby is suffering from a massive pleural effusion. Once the fluid has been drained, Bobby’s condition subsequently improves.
Actors: Respiratory Therapist, Ordering provider, attending provider, bedside terminal (scheduled/cross-over nurse), central nursing station terminal (charge nurse)
Requirements:
Requirement # | Description |
1.7.1 | The System shall have access to relevant patient medical data including diagnoses, medications, adverse events, radiology, labs, appointments, etc. required to evaluate new patient data. |
1.7.2 | The System shall provide an inference engine capable of using a variety of standard signal-averaging techniques on medical device data. Real-time signal processing is often required to reduce signal noise in streaming physiologic data, eliminating random waveform artifacts and delivering clean, accurate data to the inference engine. It is also useful to reduce the volume of data being delivered. |
1.7.3 | The System allows users to view and modify their personal information. If the user has appropriate privileges, it will allow them to search for any user and modify their demographic information. (Same as requirement 1.6.18). |