1: \begin{abstract}

2: As medical devices become more complex, they routinely collect

3: extensive and complicated data.  While classical regressions typically examine the relationship between an outcome and a vector of predictors,  

4: it becomes imperative to identify the relationship with predictors possessing

5: functional structures. In this article, we introduce a novel inference

6: procedure for examining the relationship between outcomes and large-scale functional predictors.

7: We target testing the linear hypothesis on the functional parameters 

8: under the generalized functional linear regression framework, where the

9: number of the functional parameters grows with the sample size. We develop

10: the estimation procedure for the high dimensional generalized

11: functional linear model incorporating B-spline functional

12: approximation and amenable regularization. Furthermore, 

13: we construct a procedure that is able to test the

14: local alternative hypothesis on the linear combinations of the

15: functional parameters. We establish the statistical guarantees in

16: terms of non-asymptotic convergence of the parameter estimation and

17: the oracle property and asymptotic normality of the

18: estimators. Moreover, we derive the asymptotic distribution of the test

19: statistic. We carry out intensive simulations and illustrate with a

20: new dataset from an Alzheimer’s disease magnetoencephalography study.

21: %The text of your abstract. 200 or fewer words.

22: \end{abstract}

23: