Closed Loop PID Control of Nasally Administered Oxygen using SpO2 feedback

More than 95% of death in children from preventable or easily treatable diseases happens in developing countries [1]. For children under 5 in that group, pneumonia is the leading cause of death [1]. Oxygen therapy (through a nasal cannula) is an effective therapy for children suffering from pneumonia and other respiratory infections [3]. Hypoxaemia (low blood oxygen levels) is the most prominent cause of death in children suffering from pneumonia [1]. While the WHO recommends oxygen titration to maintain a peripheral oxygen saturation (SpO2) of 90%+ and prevent hypoxaemia, current methods are manual and rarely result in a satisfactory amount of time within SpO2 target limits [2]. Moreover, oxygen is a valuable commodity in most of the world [1] and thus its use should be controlled and limited to only the necessary amount for each patient. This thesis therefore explores the modelling of the SpO2 response to changes in the flow rate of nasally administered oxygen using data collected from a study done in collaboration between the BC Children’s Research Institute and Holy Innocents Hospital, Uganda. The models are then used to inform the design of a PID controller that is tolerant to all the obtained models. This is done by performing constrained optimization over the model set, minimizing the IAE (Integrated Absolute Error) of the error signal for a load step disturbance rejection while limiting the maximum sensitivity Ms to ensure tolerance to modelling errors and inter-patient variability. Finally, the controller is simulated in different scenarios and its performance is evaluated.