The Autonomic Nervous System (ANS) regulates several physiological processes, including cardiovascular signals such as Heart Rate (HR) and Blood Pressure (BP). Several mathematical models such as WindKessel and Multivariate Autoregressive Models have been used to try to estimate these signals and their relations. Within a Point Process framework, this thesis work aimed at investigating the effects of a third covariate, the Pulse Arrival Time (PAT), on autonomic measures of interest that might be affected by vasoactivity. To this end, the thesis work has been developed in 4 phases: 1. adaptation of spectral indexes such as directional gains and coherences to a trivariate framework; 2. validation of the new trivariate indexes on surrogate data; 3. validation of the new indexes on tilt table test recordings, while evaluating new PAT related behaviours; 4. assessment of trivariate indexes in an ICU clinical context, with a focus on sepsis characterization. Starting from the first step of the definition of the new trivariate model, time-varying simulated signals with pre-defined amplitude and phase changes were used to assess its ability to identify time-varying spectral properties: the new trivariate directional gains and phases indeed identified and correctly quantified the imposed changes. Then, the new model has been used on real data from a tilt table test. The trivariate indexes kept the discriminative power of their bivariate analogues, as assessed by signed rank test on the median of tilt phases against the median of rest phases; though, plots of the trivariate median spectral indexes during tilt protocols showed interesting patterns that might be related to vasoactivity. When the model is applied to ICU sepsis recordings, the trivariate indexes results agree with the deteriorated state of the vasculature and general autonomic impairment in sepsis. In the end, the addition of PAT as a third covariate helped shed light on the state of the vasculature and provided an assessment of the influences of mechanical, autonomic or pathological nature acting on it. From the available literature, only a few papers used statistical black-box models to quantify or provide insights on vascular activity by itself. Importantly, our study was the first to use a trivariate framework including PAT to characterize Sepsis in ICU pathological data.
Il Sistema Nervoso Autonomo (SNA) regola vari processi fisiologici, inclusi segnali cardiovascolari come la frequenza cardiaca (HR) e la pressione sanguigna (BP). Molti modelli matematici come WindKessel e gli Autoregressivi Multivariati sono stati usati per stimare questi segnali e le loro relazioni. Utilizzando un modello Point Process, questo lavoro di tesi vuole valutare gli effetti di una terza covariata, il Pulse Arrival Time (PAT), su misure autonomiche di rilievo che potrebbero essere influenzate dalla vasoattività. Quindi, il lavoro di tesi è stato svolto su 4 fasi: 1. adattamento degli indici spettrali come i gain direzionali e le coerenze ad un framework trivariato; 2. validazione dei nuovi indici trivariati su dati simulati; 3. validazione dei nuovi indici su registrazioni di tilt table test, valutando la presenza di nuovi andamenti legati al PAT; 4. valutazione dei nuovi indici trivariati in contesto clinico di terapia intensiva, con particolare attenzione alla caratterizzazione della sepsi. Partendo dalla prima fase di definizione un modello trivariato, per valutarne la capacità di identificare proprietà spettrali tempo-varianti è stato testato su segnali simulati con cambi predefiniti di ampiezza e fase nel tempo: gain e fasi direzionali hanno identificato adeguatamente i cambiamenti imposti. A seguire, il nuovo modello è stato usato su dati reali da tilt table test. Gli indici trivariati hanno mantenuto la capacità discriminativa dei loro analoghi bivariati, come accertato dal signed rank test sulla mediana delle fasi tilt contro la mediana delle fasi rest; però, grafici dell'andamento temporale mediano degli indici spettrali trivariati durante i protocolli di tilt hanno mostrato pattern interessanti che potrebbero essere legati alla vasoattività. Applicando il modello ai dati dei pazienti settici in terapia intensiva, alcuni dei risultati trivariati sono associabili allo stato deteriorato della vascolatura e alla risposta autonomica depressa nella sepsi. In conclusione, l'aggiunta di una terza covariata PAT sembra aver portato alla luce lo stato della vascolatura e degli effetti esterni che la influenzano (che siano meccanici, autonomici o patologici). Nella letteratura analizzata, pochi paper riportano l'uso di modelli statistici black-box per quantificare o analizzare l'attività vascolare di per sé. L'apporto più rilevante della nostro progetto di tesi, è quello di essere il primo ad usare un modello trivariato che include PAT per caratterizzare la sepsi in un contesto di terapia intensiva.
Assessment of pulse arrival time effects in a trivariate cardiovascular model on diagnostic indexes in the ICU
MANDORLINI, MARTINA;Franchi, Eriberto Andrea
2021/2022
Abstract
The Autonomic Nervous System (ANS) regulates several physiological processes, including cardiovascular signals such as Heart Rate (HR) and Blood Pressure (BP). Several mathematical models such as WindKessel and Multivariate Autoregressive Models have been used to try to estimate these signals and their relations. Within a Point Process framework, this thesis work aimed at investigating the effects of a third covariate, the Pulse Arrival Time (PAT), on autonomic measures of interest that might be affected by vasoactivity. To this end, the thesis work has been developed in 4 phases: 1. adaptation of spectral indexes such as directional gains and coherences to a trivariate framework; 2. validation of the new trivariate indexes on surrogate data; 3. validation of the new indexes on tilt table test recordings, while evaluating new PAT related behaviours; 4. assessment of trivariate indexes in an ICU clinical context, with a focus on sepsis characterization. Starting from the first step of the definition of the new trivariate model, time-varying simulated signals with pre-defined amplitude and phase changes were used to assess its ability to identify time-varying spectral properties: the new trivariate directional gains and phases indeed identified and correctly quantified the imposed changes. Then, the new model has been used on real data from a tilt table test. The trivariate indexes kept the discriminative power of their bivariate analogues, as assessed by signed rank test on the median of tilt phases against the median of rest phases; though, plots of the trivariate median spectral indexes during tilt protocols showed interesting patterns that might be related to vasoactivity. When the model is applied to ICU sepsis recordings, the trivariate indexes results agree with the deteriorated state of the vasculature and general autonomic impairment in sepsis. In the end, the addition of PAT as a third covariate helped shed light on the state of the vasculature and provided an assessment of the influences of mechanical, autonomic or pathological nature acting on it. From the available literature, only a few papers used statistical black-box models to quantify or provide insights on vascular activity by itself. Importantly, our study was the first to use a trivariate framework including PAT to characterize Sepsis in ICU pathological data.File | Dimensione | Formato | |
---|---|---|---|
Executive Summary Franchi Mandorlini.pdf
accessibile in internet per tutti
Descrizione: Executive summary, Assessment of Pulse Arrival Time Effects in a Trivariate Cardiovascular Model on Diagnostic Indexes in the ICU. Coautori: Martina Mandorlini, Eriberto Andrea Franchi.
Dimensione
1.19 MB
Formato
Adobe PDF
|
1.19 MB | Adobe PDF | Visualizza/Apri |
Assessment of Pulse Arrival Time Effects in a Trivariate Cardiovascular Model on Diagnostic Indexes in the ICU.pdf
accessibile in internet per tutti
Descrizione: Elaborato di tesi. Coautori: Martina Mandorlini, Eriberto Andrea Franchi.
Dimensione
13.63 MB
Formato
Adobe PDF
|
13.63 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/10589/197834