Systems and methods for model-driven system integration on a ventilator comprise a modeled exhalation flow. Ventilator flow sensors contain components that are easily damaged or impacted and have high rates of failure. Knowledge failure frequency and type may help determine when to replace, clean, or calibrate a flow sensor. A modeled exhalation flow may be trained for a ventilator. Additionally, the modeled exhalation flow may be compared to a flow sensor flow to determine a specific failure. Additionally or alternatively, the modeled exhalation flow may supplement, weight, or replace a faulty flow sensor measurement. These systems and methods may assist in identifying an inaccurate flow sensor measurement and/or providing a more accurate modeled flow estimate, thus increasing accuracy in patient-ventilator interactions.