Exams text collection previous years

Pagina 1

Surname _____________________________ mark of this test final mark Firs name ____________________________ Reg. No. _____________________________ sum of the 2 parts Biomedical Engineering MSc Program Politecnico di Milano Bioengineering of the Physiological Control Systems – Prof. G. Baselli February 4th, 2017 (1st call – 2nd part) – 8 questions – max. score 16/30  Duration 50 min. No material apart these sheets  Any answer scores 2/30: 1st part 14/30 + 2nd part 16/30 = 30/30. Laude upon quality  Attention: questions under the same number (e.g., 1a, 1b, ...) are connected  Attention: severely wrong answers score a penalty of -1/30  Attention: this is the 2nd part of the 1st call written test: only valid with the 1st part /2 1 1a Suppose you decided to characterize the lung compartment by means of a non-parametric estimate (i.e., FFT based) of its Frequency Response (FR) G(). You choose the inspired CO2 concentration (partial pressure) as input PI,CO2(t), while the output is necessarily the alveolar CO2 partial pressure PA,CO2(t). What is the equation you would apply to estimate G()? G() =Cyu()/Su() Dove Cyu è il cross-spettro di ingresso e uscita (PI,CO2(t) e PA,CO2(t)) e Su è lo spettro dell’ingresso (PI,CO2(t)) How could you compute the variance of your estimate? σ2 = const. |G()| * (1-Kyu()2)/ Kyu()2 dove: Kyu()2 = |Cyu()|2/(Su() * Sy()) è detta coerenza I/O How would you exploit the latter in order to average many subjects? Farei una media pesata dove ogni soggetto è pesato per il reciproco della sua varianza (1/ σ2) così che i dati più precisi pesino di più in quanto hanno σ2 minore e quindi 1/ σ2 maggiore. /2 2 1b As in Khoo’s method (that, conversely was parametric) you can set only two PI,CO2 values at each inspiration: 40 mmHg (6%) or 0. [?? –…