Seismocardiography (SCG) is the measurement of vi-brations on the chest wall originating from the heart. Thisnon-invasive method provides information about the me-chanical components of the heart. The morphology of theSCG-signal changes in relation to respiration, and under-standing these changes will improve the diagnostic valueof SCG. The aim of this study was to determine the am-plitude of a respiration signal measured through a nasalcatheter, using SCG features, for specific time points atmitral closure (MC) and aortic opening (AO) in a three-model comparison. The three proposed methods were mul-tiple regression analysis (MRA), support vector regression(SVR), and a neural network (NN). SCG, Electrocardio-graphy (ECG) and nasal-catheter flow signals were col-lected from 18 healthy subjects (age 29+/-6 years). Be-sides MC and AO were the isovolumic moment (IM), aorticclosure (AC) and mitral opening (MO) selected as fiducialpoints in the SCG-signal. These were found using an auto-matic algorithm followed by manual verification. Fiducialpoints amplitudes, frequency components, and timings be-tween these fiducial points formed 12 features. All modelswere trained on 80% of the data and underwent 10-foldcross validation. Testing was performed on the remain-ing 20% of the data. Predictions on test data for MC andAO time points, the Pearson correlations coefficient, andsum of squared errors of prediction were: (r_MC, r_AO,SSE_MC, SSE_AO) for the following models: NN (0.908,0.904, 11.71, 12.05), SVR (0.881, 0.833, 18.95, 19.76)and MRA (0.450, 0.437, 51.21, 51.48). These predictivemodels shows a measurable correlation between the SCG-signal and respiration, which enables modelling of the cor-relation.