Personalized PEEP adjustment method based on reinforcement learning

Abstract

The invention provides a reinforcement learning-based personalized PEEP adjustment method, which comprises the following steps of: constructing a reinforcement learning model based on physiological data of a patient in the whole mechanical ventilation process, and recommending a personalized PEEP level of the patient in the next time period according to the physiological data of the patient by the reinforcement learning model to realize personalized PEEP adjustment. According to the reinforcement learning model, a Markov framework is constructed, all single-step conversion tuples in physiological data of each patient in the whole mechanical ventilation process are extracted, a fitting Q iterative algorithm is used as core training, and a strategy learned by the model finally is an action with the maximum accumulative return, so that the reinforcement learning model is obtained. According to the method, the patient can obtain better oxygenation and prognosis through the clinical physiological data of the patient, the personalized breathing machine PEEP setting value is recommended in real time, and clinical decision making of a doctor is assisted.

Description

technical field [0001] The invention relates to a personalized PEEP adjustment method based on reinforcement learning. Belongs to the technical field. Background technique [0002] Positive end-expiratory pressure (PEEP) refers to the positive pressure generated by the ventilator during the inspiratory phase during mechanical ventilation, which presses the gas into the lungs. When the airway is opened at the end of expiration, the airway pressure does not drop to 0. A certain level of positive pressure is still maintained. An appropriate PEEP value can prevent alveolar atrophy and collapse, expand trapped air bubbles and small airways, reduce alveolar and pulmonary interstitial edema, and improve lung compliance and oxygenation. It plays an important role in the treatment of acute respiratory distress syndrome (ARDS) and acute lung injury (ALI). However, paradoxically, mechanical ventilation is inherently harmful. Excessive mechanical ventilation can lead to ventilator-i...

AI Technical Summary

Problems solved by technology

There is still no clear gold standard for how to choose the optimal PEEP, and optimizing PEEP in mechanically ventilated patients remains a challenge for clinicians

Images