Keeping the Patient in the Loop

Closed-loop control systems, which adjust device output based on information received from a sensor to keep a variable at a reference position, are common in many medical devices.  There are numerous examples where device output is controlled to maintain a physical measurement, such as pressure, at a set point.  When the variable of interest is a physiologic measurement, the patient becomes part of the closed-loop control system and clinician involvement in responding to changes in the patient’s condition can be reduced, leading to the emergence of new types of risks.

On December 23, 2021, CDRH released a draft guidance document, Technical Considerations for Medical Devices with Physiologic Closed-Loop Control Technology (PCLC Draft Guidance) that describes design, testing and labeling considerations to characterize and control the unique risks associated with physiological closed-loop controlled (PCLC) devices.  The PCLC Draft Guidance defines a PCLC device as a system consisting of physiologic-measuring sensors, actuators, and control algorithms that adjusts or maintains a physiologic variable (e.g., mean arterial blood pressure, depth of anesthesia) through automatic adjustments to delivery or removal of energy or article (e.g., drugs, or liquid or gas regulated as a medical device) using feedback from a physiologic-measuring sensor(s).  PCLC Draft Guidance at 4.

Premarket applications for PCLC devices should describe the PCLC device using functional block diagrams and provide descriptive content on control algorithms, sensors, user interface and system safety features.  Safety features can include:  fallback modes that the device enters when unsafe conditions are detected; transparent entrance and exit criteria for initiation and cessation of automated therapy; constraints on delivered energy or article, such as upper/lower limits and total amounts delivered over time; data logging; and alarms.  The PCLC Draft Guidance recommends that patient-related hazards, especially inter- and intra-patient variability, device-related hazards, and use-related hazards be evaluated as part of the risk analysis.

We found the discussion on testing of PCLC devices the most interesting.  A PCLC device will need to be tested via a broad range of assessments, including those common to many device types, with additional testing specific to PCLC devices. The PCLC Draft Guidance provides specific recommendations related to animal testing, testing using mathematical and computational models and human factors testing.  We were interested to see that clinical testing does not appear to be a major focus for these devices, though it is mentioned as a possible means of validation.  Given the complexities of PCLC devices, the PCLC Draft Guidance recommends use of the pre-submission process to receive Agency feedback, especially on animal test protocols, use of mathematical models and human factors testing.

The PCLC Draft Guidance recommends verification of PCLC devices include demonstration that sensors, actuators, and safety features meet all specifications and that the PCLC system response meets specifications during normal and foreseeable worst-case conditions and during foreseeable functional and clinical disturbances.  Parameter sensitivity analysis can be performed to demonstrate that the device meets specifications across all combinations of adjustable parameter values.  Validation specific to PCLC devices should cover user interactions and demonstrate that the PCLC device performs as intended and that its response supports safe and effective operation during normal and foreseeable worst-case conditions.  Id. at 21-22.

Entirely virtual testing refers to testing that is performed completely in a simulated computer environment.  Id. at 26.  Hardware-in-the-loop testing is performed using computational models of the patient’s physiology interfaced with the PCLC device hardware.  Id. at 27.  Credibility of models used in these types of assessments should be evaluated according to the draft guidance, Assessing the Credibility of Computational Modeling and Simulation in Medical Device Submissions, which we blogged about here.  Both of these types of assessments using computational models allow for simulation across a wide range of scenarios including inter-patient and intra-patient variability and uncertainty.  Inclusion of device hardware in hardware in-the-loop testing can be useful in identifying system failure modes and hardware limitations.

The PCLC Draft Guidance emphasizes the need for a robust user training program that is incorporated into human factors evaluations.  Training should be developed such that trainees experience complacency, automation bias, and loss of situational awareness, which are all risks introduced with PCLC technology.  Training should also include automation failure to give users practice responding to these situations.  Id. at 29.  Human factors testing can include simulated or actual use testing.  Because automation-related use error might not be predictable, human factors testing conducted in a clinical setting is recommended to enable realistic and meaningful evaluation. Id. at 28.

Overall, the PCLC Draft Guidance provides important considerations for the design and testing of devices incorporating PCLC technology that should benefit development of these devices and future interactions with the Agency.