During a surgical operation, the patient's anesthetic status is a crucial factor for the success of the operation. In this context the anesthesiologist has the role of identifying the dose of drug to be administered and the corrections necessary during the intervention, based exclusively on his observations and experience. The administration dose should guarantee the correct state of anesthesia and at the same time minimize undesirable effects by improving post-operative recovery. The research is therefore aimed to develop instruments to efficiently control the infusion of the drug, and on the other hand to identify pharmacokinetic-pharmacodynamic models (PK-PD) that describe the dose-effect relationship keeping into account the physiological characteristics of the patient. The aim of the present thesis work is twofold: initially obtain a PK-PD model that describes the behavior of propofol and ketamine in the feline species and subsequently develop a control system for the automated administration of the two drugs (to support the veterinarian anesthesiologist). Regarding the PK-PD model, we based our work on the Physiologically-Based Pharmacokinetic model PBPK proposed by Abbiati et al. (2016), adapting it to the feline physiology and to the chemical-physical characteristics of the studied drugs. Pharmacodynamics has been described by the modified Hill equation. The introduction of an effect-site compartment in the PK model allowed to model the delay that exists between the plasma concentration and the occurrence of the pharmacological effect. The model has been identified and validated using clinical data provided by the Department of Veterinary Sciences at the University of Milan. The automated anesthesia control was entirely in-silico simulated: the control loop consists of a virtual patient, whose physiological variables such as blood pressure and heart rate are simulated by the PBPK-PD model, and by a controller, whose control actions are based on the prediction of the future evolution of the patient's state through a model ("Model Predictive Control"). In order to simulate the discrepancy (existing in the reality between real patient and its mathematical model) between the virtual patient and the model of the controller ("model mismatch"), the latter has been described through a simplified PK-PD model. This system was tested by simulating in-silico two different stages of anesthesia: (i) the induction of anesthesia/analgesia (servo problem) and (ii) the maintenance of the anesthetic state in the presence of external disturbances (regulator problem). It has been proved that the analyzed control strategy is able to effectively address the induction phase of anesthesia, allowing to reduce the total amount of drug administered, ensuring reduced control efforts and limiting the increase in plasma concentration, although maintaining a satisfactory reactivity. In the case of ketamine, the controller is able to suppress the disturbance effectively and quickly. In the case of propofol, although the controller is able to bring the HR below the set point value very quickly, the performances obtained are not optimal.
Durante un'operazione chirurgica, lo stato anestetico del paziente è un fattore cruciale per la buona riuscita dell’operazione stessa. In questo contesto il medico anestesista ha il ruolo di individuare la dose di farmaco da somministrare e le correzioni necessarie durante l’intervento, basandosi esclusivamente sulle proprie osservazioni ed esperienza. La somministrazione della dose dovrebbe garantire il corretto stato di anestesia e al contempo minimizzare gli effetti indesiderati migliorando la ripresa post-operatoria. La ricerca è quindi volta da un lato allo sviluppo di strumenti che controllino in modo regolare ed efficiente l’infusione del farmaco, e dall’altro all’individuazione di modelli farmacocinetici-farmacodinamici (PK-PD) che descrivano la relazione dose-effetto tenendo conto delle caratteristiche fisiologiche del paziente. Lo scopo del presente lavoro di tesi è duplice: inizialmente ottenere un modello PK-PD che descriva il comportamento di propofol e ketamina nella specie felina e successivamente sviluppare un sistema di controllo per la somministrazione automatizzata dei due farmaci (a supporto del medico/veterinario anestesista). Per quanto riguarda il modello PK-PD, ci si è basati sul modello farmacocinetico basato su fisiologia (“Physiologically-Based Pharmacokinetics”, PBPK) proposto da Abbiati et al. (2016), adattandolo alla fisiologia felina e alle caratteristiche chimico-fisiche dei farmaci studiati. La farmacodinamica è stata descritta sulla base dell’equazione di Hill. L’introduzione di un compartimento effect-site nel modello PK ha permesso di modellare il ritardo che sussiste tra la concentrazione plasmatica e il manifestarsi dell’effetto farmacologico. Il modello è stato identificato e convalidato utilizzando dati clinici forniti dal Dipartimento di Scienze Veterinarie presso l’Università degli Studi di Milano. Il controllo automatizzato dell’anestesia è stato interamente simulato in-silico: l’anello di controllo è costituito da un paziente virtuale, le cui variabili fisiologiche come pressione arteriosa e frequenza cardiaca sono simulate dal modello PBPK-PD, e da un controllore, le cui azioni di controllo si basano sulla previsione dell’evoluzione futura dello stato del paziente mediante un modello (“Model Predictive Control”). Al fine di simulare la discrepanza (esistente nella realtà tra paziente reale e modello matematico dello stesso) tra il paziente virtuale e il modello del controllore (“model mismatch”), quest’ultimo è stato descritto tramite un modello PK-PD semplificato. Il sistema così strutturato è stato testato simulando in-silico due diverse fasi dell’anestesia: (i) l’induzione dell’anestesia/analgesia (problema di servomeccanismo) e (ii) il mantenimento dello stato anestetico in presenza di disturbi esterni (problema di regolazione). Si è dimostrato che la strategia di controllo analizzata è in grado di affrontare in modo efficace la fase di induzione dell’anestesia, permettendo di ridurre la quantità totale di farmaco somministrata, garantendo sforzi di controllo ridotti e contenendo l’innalzamento della concentrazione plasmatica, pur mantenendo una soddisfacente rapidità di risposta. Nel caso della ketamina, il controllore è in grado di sopprimere in modo efficace e rapido il disturbo. Nel caso del propofol, nonostante il controllore sia in grado di riportare l’HR al di sotto del valore di set point molto rapidamente, le performance ottenute non sono ottimali.
Controllo multivariabile in silico basato su modello predittivo applicato all'anestesia veterinaria
de GIORGI, FEDERICA;CORBETTA, MANUEL
2017/2018
Abstract
During a surgical operation, the patient's anesthetic status is a crucial factor for the success of the operation. In this context the anesthesiologist has the role of identifying the dose of drug to be administered and the corrections necessary during the intervention, based exclusively on his observations and experience. The administration dose should guarantee the correct state of anesthesia and at the same time minimize undesirable effects by improving post-operative recovery. The research is therefore aimed to develop instruments to efficiently control the infusion of the drug, and on the other hand to identify pharmacokinetic-pharmacodynamic models (PK-PD) that describe the dose-effect relationship keeping into account the physiological characteristics of the patient. The aim of the present thesis work is twofold: initially obtain a PK-PD model that describes the behavior of propofol and ketamine in the feline species and subsequently develop a control system for the automated administration of the two drugs (to support the veterinarian anesthesiologist). Regarding the PK-PD model, we based our work on the Physiologically-Based Pharmacokinetic model PBPK proposed by Abbiati et al. (2016), adapting it to the feline physiology and to the chemical-physical characteristics of the studied drugs. Pharmacodynamics has been described by the modified Hill equation. The introduction of an effect-site compartment in the PK model allowed to model the delay that exists between the plasma concentration and the occurrence of the pharmacological effect. The model has been identified and validated using clinical data provided by the Department of Veterinary Sciences at the University of Milan. The automated anesthesia control was entirely in-silico simulated: the control loop consists of a virtual patient, whose physiological variables such as blood pressure and heart rate are simulated by the PBPK-PD model, and by a controller, whose control actions are based on the prediction of the future evolution of the patient's state through a model ("Model Predictive Control"). In order to simulate the discrepancy (existing in the reality between real patient and its mathematical model) between the virtual patient and the model of the controller ("model mismatch"), the latter has been described through a simplified PK-PD model. This system was tested by simulating in-silico two different stages of anesthesia: (i) the induction of anesthesia/analgesia (servo problem) and (ii) the maintenance of the anesthetic state in the presence of external disturbances (regulator problem). It has been proved that the analyzed control strategy is able to effectively address the induction phase of anesthesia, allowing to reduce the total amount of drug administered, ensuring reduced control efforts and limiting the increase in plasma concentration, although maintaining a satisfactory reactivity. In the case of ketamine, the controller is able to suppress the disturbance effectively and quickly. In the case of propofol, although the controller is able to bring the HR below the set point value very quickly, the performances obtained are not optimal.File | Dimensione | Formato | |
---|---|---|---|
2018_10_Corbetta_deGiorgi.pdf
non accessibile
Descrizione: Tesi
Dimensione
5.84 MB
Formato
Adobe PDF
|
5.84 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/142751