Pulmonary Function Tests are important non-invasive

\r\ndiagnostic tests to assess respiratory impairments and provides

\r\nquantifiable measures of lung function. Spirometry is the most

\r\nfrequently used measure of lung function and plays an essential role

\r\nin the diagnosis and management of pulmonary diseases. However,

\r\nthe test requires considerable patient effort and cooperation,

\r\nmarkedly related to the age of patients resulting in incomplete data

\r\nsets. This paper presents, a nonlinear model built using Multivariate

\r\nadaptive regression splines and Random forest regression model to

\r\npredict the missing spirometric features. Random forest based feature

\r\nselection is used to enhance both the generalization capability and the

\r\nmodel interpretability. In the present study, flow-volume data are

\r\nrecorded for N= 198 subjects. The ranked order of feature importance

\r\nindex calculated by the random forests model shows that the

\r\nspirometric features FVC, FEF25, PEF, FEF25-75, FEF50 and the

\r\ndemographic parameter height are the important descriptors. A

\r\ncomparison of performance assessment of both models prove that, the

\r\nprediction ability of MARS with the `top two ranked features namely

\r\nthe FVC and FEF25 is higher, yielding a model fit of R2= 0.96 and

\r\nR2= 0.99 for normal and abnormal subjects. The Root Mean Square

\r\nError analysis of the RF model and the MARS model also shows that

\r\nthe latter is capable of predicting the missing values of FEV1 with a

\r\nnotably lower error value of 0.0191 (normal subjects) and 0.0106

\r\n(abnormal subjects) with the aforementioned input features. It is

\r\nconcluded that combining feature selection with a prediction model

\r\nprovides a minimum subset of predominant features to train the

\r\nmodel, as well as yielding better prediction performance. This

\r\nanalysis can assist clinicians with a intelligence support system in the

\r\nmedical diagnosis and improvement of clinical care.<\/p>\r\n","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 100, 2015"}