In this study, an innovative least-squares adaptive estimator is developed for heart fault diagnosis. The reference model of this study is the one developed by Ursino in 1998. Mitral valve function is studied in this paper based on a new method which uses invasive noisy blood pressure waveform observations of left ventricle, left atrium, and left pulmonary vein. To meet this end, an adaptive algorithm is designed for estimation of discontinuous time-variant parameter values and then Ursinopsilas model simulator is utilized for the evaluation of Mitral valve non-linear gain. The results obtained, indicate the high capability of the presented model in the estimation of different cardiovascular parameters and so fault diagnosis in heart valves.