Skip to main content

Advertisement

Log in

Semiparametric partially linear varying coefficient models with panel count data

  • Published:
Lifetime Data Analysis Aims and scope Submit manuscript

Abstract

This paper studies semiparametric regression analysis of panel count data, which arise naturally when recurrent events are considered. Such data frequently occur in medical follow-up studies and reliability experiments, for example. To explore the nonlinear interactions between covariates, we propose a class of partially linear models with possibly varying coefficients for the mean function of the counting processes with panel count data. The functional coefficients are estimated by B-spline function approximations. The estimation procedures are based on maximum pseudo-likelihood and likelihood approaches and they are easy to implement. The asymptotic properties of the resulting estimators are established, and their finite-sample performance is assessed by Monte Carlo simulation studies. We also demonstrate the value of the proposed method by the analysis of a cancer data set, where the new modeling approach provides more comprehensive information than the usual proportional mean model.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

References

  • Andrews DF, Herzberg AM (1985) Data: a collection of problems from many fields for the student and research worker. Springer, New York

    Book  MATH  Google Scholar 

  • Balakrishnan N, Zhao X (2009) New multi-sample nonparametric tests for panel count data. Ann Stat 37:1112–1149

    Article  MathSciNet  MATH  Google Scholar 

  • Byar DP (1980) The Veterans Administration study of chemoprophylaxis for recurrent stage I bladder tumors: comparisons of placebo, pyridoxine, and topical thiotepa. In: Pavane-Macaluso M, Smith PH, Edsmyr F (eds) Bladder tumors and other topics in urological oncology. Plenum, New York, pp 363–370

    Chapter  Google Scholar 

  • Cheng MY, Honda T, Li J, Peng H (2014) Nonparametric independence screening and structure identification for ultra-high dimensional longitudinal data. Ann Stat 42:1819–1849

    Article  MathSciNet  MATH  Google Scholar 

  • Cheng SC, Wei LJ (2000) Inferences for a semiparametric model with panel data. Biometrika 87:89–97

    Article  MathSciNet  MATH  Google Scholar 

  • Hu XJ, Lagakos SW, Lockhart RA (2009a) Generalized least squares estimation of the mean function of a counting process based on panel counts. Stat Sinica 19:561–580

    MathSciNet  MATH  Google Scholar 

  • Hu XJ, Lagakos SW, Lockhart RA (2009b) Marginal analysis of panel counts through estimating functions. Biometrika 96:445–456

    Article  MathSciNet  MATH  Google Scholar 

  • Hu XJ, Sun J, Wei LJ (2003) Regression parameter estimation from panel counts. Scand J Stat 30:25–43

    Article  MathSciNet  MATH  Google Scholar 

  • Hua L, Zhang Y (2012) Spline-based semiparametric projected generalized estimating equation method for panel count data. Biostatistics 13:440–454

    Article  MATH  Google Scholar 

  • Huang J (1999) Efficient estimation of the partly linear additive Cox model. Ann Stat 27:1536–1563

    Article  MathSciNet  MATH  Google Scholar 

  • Huang CY, Wang MC, Zhang Y (2006) Analysing panel count data with informative observation times. Biometrika 93:763–775

    Article  MathSciNet  MATH  Google Scholar 

  • Lindqvist BH, Kvaløy JT, Aaserud S (2015) Residual plots to reveal the functional form for covariates in parametric accelerated failure time models. Lifetime Data Anal 21:353–378

    Article  MathSciNet  MATH  Google Scholar 

  • Lu M, Zhang Y, Huang J (2007) Estimation of the mean function with panel count data using monotone polynomial splines. Biometrika 94:705–718

    Article  MathSciNet  MATH  Google Scholar 

  • Lu M, Zhang Y, Huang J (2009) Semiparametric estimation methods for panel count data using monotone B-splines. J Am Stat Assoc 104:1060–1070

    Article  MathSciNet  MATH  Google Scholar 

  • Schumaker L (1981) Spline functions: basic theory. Wiley, New York

    MATH  Google Scholar 

  • Sun J, Fang HB (2003) A nonparametric test for panel count data. Biometrika 90:199–208

    Article  MathSciNet  MATH  Google Scholar 

  • Sun J, Kalbfleisch JD (1995) Estimation of the mean function of point processes based on panel count data. Stat Sinica 5:279–290

    MathSciNet  MATH  Google Scholar 

  • Sun J, Tong X, He X (2007) Regression analysis of panel count data with dependent observation times. Biometrics 63:1053–1059

    Article  MathSciNet  MATH  Google Scholar 

  • Sun J, Wei LJ (2000) Regression analysis of panel count data with covariate-dependent observation and censoring times. J R Stat Soc Ser B 62:293–302

    Article  MathSciNet  Google Scholar 

  • Thall PF, Lachin JM (1988) Analysis of recurrent events: nonparametric methods for random-interval count data. J Am Stat Assoc 83:339–347

    Article  Google Scholar 

  • Van der Vaart AW (2002) Semiparametric Statistics. In Lectures on Probability Theory and Statistics, Ecole d’Eté de Probabilités de Saint-Flour XXIX-1999. Lecture Notes in Mathematics 1781:330–457

  • Van der Vaart AW, Wellner JA (1996) Weak convergence and empirical processes. Springer, New York

    Book  MATH  Google Scholar 

  • Wellner JA, Zhang Y (2000) Two estimators of the mean of a counting process with panel count data. Ann Stat 28:779–814

    Article  MathSciNet  MATH  Google Scholar 

  • Wellner JA, Zhang Y (2007) Two likelihood-based semiparametric estimation methods for panel count data with covariates. Ann Stat 35:2106–2142

    Article  MathSciNet  MATH  Google Scholar 

  • Zeng D, Lin D, Yin GS (2005) Maximum likelihood estimation for the proportional odds model with random effects. J Am Stat Assoc 100:470–482

    Article  MathSciNet  MATH  Google Scholar 

  • Zhang Y (2002) A semiparametric pseudolikelihood estimation method for panel count data. Biometrika 89:39–48

    Article  MathSciNet  MATH  Google Scholar 

  • Zhang Y (2006) Nonparametric \(k\)-sample tests for panel count data. Biometrika 93:777–790

    Article  MathSciNet  MATH  Google Scholar 

  • Zhang F, Chen X, Zhou Y (2014) Proportional hazards model with varying coefficients for length-biased data. Lifetime Data Anal 20:132–157

    Article  MathSciNet  MATH  Google Scholar 

  • Zhang Y, Jamshidian M (2003) The gamma-frailty Poisson model for the nonparametric estimation of panel count data. Biometrics 59:1099–1106

    Article  MathSciNet  MATH  Google Scholar 

Download references

Acknowledgments

The authors would like to thank the Editor, Professor Mei-Ling Ting Lee, the Associate Editor and the two reviewers for their constructive and insightful comments and suggestions that greatly improved the paper. Tong’s research was partly supported by BCMIIS, NSF China Zhongdian Project 11131002 and NSFC (No. 10971015). Zhao’s research was partly supported by the Research Grants Council of Hong Kong (PolyU 504011 and PolyU 503513), the Natural Science Foundation of China (No. 11371299), and The Hong Kong Polytechnic University.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Xingqiu Zhao.

Appendix: Proofs of asymptotic results

Appendix: Proofs of asymptotic results

In this section we present the proofs of Theorems 3.13.2 and 3.3.

1.1 Proof of Theorem 3.1

Here, we only present the proof of part (i) since part (ii) can be verified similarly. Let

$$\begin{aligned} m^{ps}_\theta (X)= & {} \sum ^{K}_{j=1}\left[ N(T_{K,j})\log \{\Lambda (T_{K,j})\exp {(\beta 'Z+V\phi (W))}\}\right. \\&\left. -\Lambda (T_{K,j})\exp {(\beta 'Z+V\phi (W))} \right] ,\\ M^{ps}_n(\theta )= & {} P_nm^{ps}_\theta (X), \quad \text{ and } \quad M^{ps}(\theta )=Pm^{ps}_\theta (X), \end{aligned}$$

where P and \(P_n\) denote the probability measure and the empirical measure, respectively. Let \(h(x)=x\log x-x+1\). Note that \(h(x)\ge (x-1)^2/4\) for \(0\le x\le 5\). For any \(\theta \) in a sufficiently small neighborhood of \(\theta _0\),

$$\begin{aligned}&M^p(\theta _0)-M^p(\theta ) \nonumber \\&\quad = \int \Lambda (u)\exp \{Z'\beta +v\phi (w)\} h\left( \frac{\Lambda _0(u)\exp (z'\beta _0+v\phi _0(w))}{\Lambda (u)\exp (z'\beta +v\phi (w))}\right) d\nu _1(u, z, v,w)\nonumber \\&\quad \ge \frac{1}{4}\int \Lambda (u)\exp \{Z'\beta +v\phi (w)\} \left\{ \frac{\Lambda _0(u)\exp (z'\beta _0+v\phi _0(w))}{\Lambda (u)\exp (z'\beta +v\phi (w))}-1\right\} ^2d\nu _1(u, z, v,w).\nonumber \\ \end{aligned}$$
(7.1)

Then, using (7.1) and the arguments similar to those in Wellner and Zhang (2007), we can show that \(M^{ps}(\theta _0)=M^{ps}(\theta )\) if and only if \(\beta =\beta _0\), \(\Lambda (t)=\Lambda _0(t)\) a.e. with respect to \(\mu _1\), and \(\phi =\phi _0\) by C3 and C7.

By the similar arguments as those used in Wellner and Zhang (2007) again, we can also show that \({\hat{\Lambda }}^{ps}_n(t)\) is uniformly bounded in probability for \(t\in [0, b]\) if \(\mu _1([b, \tau ])>0\) for some \(0<b<\tau \) or \(t\in [0, \tau ]\) if \(\mu _1(\{\tau \})>0\).

By Helly-Selection Theorem and compactness of \(\Theta _n\), it follows that \({\hat{\theta }}_n^{ps}=({\hat{\beta }}^{ps}_n, {\hat{\Lambda }}^{ps}_n, {\hat{\phi }}^{ps}_n)\) has a subsequence \({\hat{\theta }}^{ps}_{n_k}=({\hat{\beta }}^{ps}_{n_k}, {\hat{\Lambda }}^{ps}_{n_k}, {\hat{\phi }}^{ps}_{n_k})\) converging to \(\theta ^+=(\beta ^+, \Lambda ^+, \phi ^+)\), where \(\Lambda ^+\) is a nondecreasing bound function on [0, b] for \(0<b<\tau \) and it can be defined on \([0, \tau ]\) if \(\mu _1(\{\tau \})>0\).

Note that \(\Theta _n\) is compact, and the function \(m_\theta ^{ps}(x)\) is upper semicontinuous in \(\theta \) for almost all x. Furthermore, \(m^{ps}_\theta (X)\le M^{ps}_0(X)<\infty \) with \(PM^{ps}_0(X)<\infty \) by C4. Thus, by Theorem A.1 of Wellner and Zhang (2000), we have

$$\begin{aligned} \limsup _{n\rightarrow \infty }\sup _{\theta \in \Theta _n} (P_n-P)m^{ps}_\theta (X)\le 0 \qquad a.s. \end{aligned}$$
(7.2)

By the Dominated Convergence Theorem and C4, \(M^{ps}(\theta )\) is continuous in \(\theta \). Therefore, for any \(\varepsilon >0\), there exists \(\phi ^*_0\in \Psi _n\) such that

$$\begin{aligned} M^{ps}(\beta _0, \Lambda _0, \phi _0)-\varepsilon \le M^{ps}(\beta _0, \Lambda _0, \phi ^*_0) \quad \text{ with } \; ||\phi _0-\phi _0^*||_{\infty }=o(1). \end{aligned}$$

Clearly,

$$\begin{aligned} M_n^{ps}(\beta _0, \Lambda _0, \phi _0^*)-M^{ps}(\beta _0, \Lambda _0, \phi _0^*)=o_p(1) \end{aligned}$$

and

$$\begin{aligned} M_n^{ps}(\beta _0, \Lambda _0, \phi _0^*)\le M_n^{ps}({\hat{\beta }}^{ps}_n, {\hat{\Lambda }}_n^{ps}, {\hat{\phi }}_n^{ps}). \end{aligned}$$

Then, using (7.2) and the arguments similar to those used in Lu et al. (2009), we can show that \(M^{ps}(\theta ^+)=M^{ps}(\theta _0)\), which yields \(\beta ^+=\beta _0\), \(\Lambda ^+=\Lambda _0\), a.e., and \(\phi ^+=\phi _0\). Therefore, we obtain the weak consistency of \(({\hat{\beta }}_n^{ps}, {\hat{\Lambda }}^{ps}_n, {\hat{\phi }}^{ps}_n)\) in the metric \(d_1\).

1.2 Proof of Theorem 3.2

To obtain the rate of convergence, we will apply Theorem 3.2.5 of Van der Vaart and Wellner (1996). Let \( m^{ps}_\theta (X)\), \( M^{ps}_n(\theta )\), and \(M^{ps}(\theta )\) be as defined in Appendix A.1. Let \(\mu (u,v,w)=\Lambda (u)\exp \{v\phi (w)\}\), \(\mu _0(u,v,w)=\Lambda _0(u)\exp \{v\phi _0(w)\}\) and \(g(t)=\mu _t(U, Z, V,W)\exp (Z'\beta _t)\), where \((U,Z,V,W)\sim \nu _1\), \(\mu _t=t\mu +(1-t)\mu _0\), \(\beta _t=t\beta +(1-t)\beta _0\) for \(0\le t\le 1\). Then,

$$\begin{aligned} \Lambda (U)e^{Z'\beta +V\phi (W)}-\Lambda _0(U)e^{Z'\beta _0+V\phi _0(W)}=g(1)-g(0). \end{aligned}$$

By the mean value theorem, there exits a \(0\le \xi \le 1\) such that \(g(1)-g(0)=g'(\xi )\). Since

$$\begin{aligned} g'(\xi )= & {} \exp (Z'\beta _\xi )[(\mu -\mu _0)(U,V,W)+\{\mu _0+\xi (\mu -\mu _0)\}(U,V,W)Z'(\beta -\beta _0)]\\= & {} \exp (Z'\beta _\xi )[(\mu -\mu _0)(U,V,W)\{1+\xi Z'(\beta -\beta _0)\}\\&+\, \mu _0(U,V,W)Z'(\beta -\beta _0)], \end{aligned}$$

then from (7.1) we have

$$\begin{aligned}&M^{ps}(\theta _0)-M^{ps}(\theta )\\&\quad \ge c_1\int \left\{ \Lambda (u)\exp (z'\beta +v\phi (w))-\Lambda _0(u)\exp (z'\beta _0+v\phi _0(w))\right\} ^2d\nu _1(u,z,v,w)\\&\quad = c_1 \int [(\mu -\mu _0)(u,v,w)\{1+\xi z'(\beta -\beta _0)\}\\&\qquad +\, \mu _0(u,v,w)z'(\beta -\beta _0)]^2 d\nu _1(u,z,v,w)\\&\quad =c_1\nu _1(g_1h_1+g_2)^2 \end{aligned}$$

for a constant \(c_1\), where \(g_1(U,Z,V,W)=Z'(\beta -\beta _0)\mu _0(U,V,W)\), \(g_2(U,V,W)=(\mu -\mu _0)(U,V,W)\), and \(h_1(U,Z,V,W)=1+\xi \frac{(\mu -\mu _0)(U,V,W)}{\mu _0(U,V,W)}\) in the notation of Lemma 8.8 of van der Vaart (2002, page 432). To apply van der Vaart’s Lemma, we need to show that

$$\begin{aligned} \{\nu _1(g_1g_2\}^2\le c\nu _1(g_1^2)\nu _1(g_2^2) \end{aligned}$$
(7.3)

for a constant \(c<1\). By the Cauchy-Schwarz inequality and condition (C13), we can show that (7.3) holds for \(c=1-\eta _1\). Let

$$\begin{aligned} \Lambda _t=t\Lambda +(1-t)\Lambda _0, \phi _t=t\phi +(1-t)\phi _0, Q(t)=\Lambda _t(U)e^{V\phi _t(W)}. \end{aligned}$$

Then

$$\begin{aligned} g_2(U,V,W)=Q(1)-Q(0)=Q'(\zeta ) \quad \text{ for } \; 0\le \zeta \le 1, \end{aligned}$$

and

$$\begin{aligned} \nu _1(g_2^2)=\nu _1((h_2g_3+g_4)^2) \end{aligned}$$

where \(g_3(U,V,W)=V(\phi (W)-\phi _0(W))\Lambda _0(U)\), \(g_4(U)=(\Lambda -\Lambda _0)(U)\), and \(h_2(U,V,W)=1+\zeta \frac{(\Lambda -\Lambda _0)(U)}{\Lambda _0(U)}\). Similarly, we can show that

$$\begin{aligned} \{\nu _1(g_3g_4\}^2\le (1-\eta _2)\nu _1(g_3^2)\nu _1(g_4^2). \end{aligned}$$

So, by van der Vaart’s Lemma, we have

$$\begin{aligned} \nu _1(g_1h+g_2)^2 \ge cd^2_1(\theta , \theta _0). \end{aligned}$$

To derive the rate of convergence, next we need to find a \(\varphi _n(\delta )\) such that

$$\begin{aligned} E\left\{ \sup _{d_1(\theta , \theta _0)<\delta }\sqrt{n}|(P_n-P)(m^{ps}_\theta (X)-m^{ps}_{\theta _0}(X))|\right\} \le c\varphi _n(\delta ). \end{aligned}$$

Let

$$\begin{aligned} \mathcal{F}^{ps}_\delta =\left\{ m^{ps}_\theta (X)-m^{ps}_{\theta _0}(X): d_1(\theta , \theta _0)\le \delta \right\} . \end{aligned}$$

From the result of Theorem 2.7.5 of Van der Vaart and Wellner (1996) and Lemma A.2 of Huang (1999), for any \(\epsilon \le \delta \), we have

$$\begin{aligned} \log N_{[]}(\epsilon , \mathcal{F}^{ps}_\delta , ||\cdot ||_{P, B})\le c\left( \frac{1}{\epsilon }+q_n\log \frac{\delta }{\epsilon }\right) , \end{aligned}$$

where \(||\cdot ||_{P, B}\) is the Bernstein Norm defined as \(||f||_{P, B}=\{2P(e^{|f|}-1-|f|)\}^{1/2}\) by van der Vaart and Wellner (1996, page 324). Moreover, we can show that

$$\begin{aligned} ||m^{ps}_\theta (X)-m^{ps}_{\theta _0}(X)||^2_{P, B}\le c\delta ^2, \end{aligned}$$

for any \(m^{ps}_\theta (X)-m^{ps}_{\theta _0}(X)\in \mathcal{F}^{ps}_\delta \). Therefore, by Lemma 3.4.3 of Van der Vaart and Wellner (1996), we obtain

$$\begin{aligned} E||n^{1/2}(P_n-P)||_{\mathcal{F}^{ps}_\delta }\le cJ_{[ ]}(\delta ,\mathcal{F}^{ps}_\delta , ||\cdot ||_{P, B} ) \left\{ 1+\frac{J_{[ ]}(\delta ,\mathcal{F}^{ps}_\delta , ||\cdot ||_{P, B} )}{\delta ^2 n^{1/2}}\right\} \end{aligned}$$

where

$$\begin{aligned} J_{[ ]}(\delta ,\mathcal{F}^{ps}_\delta , ||\cdot ||_{P, B} )= & {} \int ^\delta _0 \{1+\log N_{[]}(\epsilon , \mathcal{F}^{ps}_\delta , ||\cdot ||_{P, B} )\}^{1/2}d\epsilon \\\le & {} c q_n^{\frac{1}{2}}\int ^\delta _0\left( 1+\frac{1}{\epsilon }+\log \frac{\delta }{\epsilon }\right) ^{1/2}d\epsilon \le c q_n^{\frac{1}{2}}\delta ^{\frac{1}{2}}. \end{aligned}$$

Thus,

$$\begin{aligned} \varphi _n(\delta )=c q_n^{\frac{1}{2}}\delta ^{\frac{1}{2}}\left( 1+\frac{cq_n^{1/2}\delta ^{1/2}}{\delta ^2n^{1/2}}\right) =c(q_n^{\frac{1}{2}}\delta ^{\frac{1}{2}}+\frac{q_n}{\delta n^{1/2}}). \end{aligned}$$

It is easy to see that \(\varphi _n(\delta )/\delta \) is decreasing in \(\delta \), and

$$\begin{aligned} r_n^2\varphi _n\left( \frac{1}{r_n}\right) =r_n^2\left( q_n^{\frac{1}{2}}r_n^{-\frac{1}{2}}+\frac{q_n}{r_n^{-1} n^{1/2}}\right) =r_n^{\frac{3}{2}}q_n^{\frac{1}{2}}+r_n^3 q_n n^{-\frac{1}{2}}\le c n^{\frac{1}{2}} \end{aligned}$$

for \(r_n=n^{\frac{1-v}{3}}\) and \(0<v<1/2\). Hence, it follows from Theorem 3.2.5 of Van der Vaart and Wellner (1996) that \( n^{\frac{1-v}{3}} d_1({\hat{\theta }}^{ps}_n, \theta _0)=O_p(1). \) Similarly, we can obtain the rate of convergence for \({\hat{\theta }}_n\).

1.3 Proof of Theorem 3.3

First, we prove part (i). Recall that

$$\begin{aligned} l^{ps}_n(\beta , \Lambda ,\phi )= & {} \sum ^n_{i=1}\sum ^{K_i}_{j=1}\left[ N_i(T_{K_i,j}) \log \left\{ \Lambda (T_{K_i,j})\right\} +N_i(T_{K_i, j})\{Z_i'\beta +V_i\phi (W_i)\} \right. \\&\left. -\Lambda (T_{K_i, j})\exp \{Z_i'\beta +V_i\phi (W_i)\}\right] . \end{aligned}$$

We define a sequence of maps \(S^{ps}_n\) mapping a neighborhood of \((\beta _0,\Lambda _0, \phi _0)\), denoted by \(\mathcal {U}\), in the parameter space for \((\beta ,\Lambda , \phi )\) into \(l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) as :

$$\begin{aligned}&S^{ps}_n(\theta )[\mathbf h _1,{h}_2, h_3]\\&\quad =n^{-1} \frac{d}{d\varepsilon } l^{ps}_n( \beta + \varepsilon \mathbf h _1,\Lambda + \varepsilon {h}_2 , \phi + \varepsilon h_3)\Big |_{\varepsilon =0}\\&\quad =n^{-1} \sum ^n_{i=1}\sum ^{K_i}_{j=1}\left[ \{N_i(T_{K_i,j})-\Lambda (T_{K_i,j})\exp (\beta 'Z_i+V_i\phi (W_i))\}{} \mathbf h '_1Z_i\right. \\&\qquad \qquad \qquad +\left\{ \frac{N_i(T_{K_i,j})}{\Lambda (T_{K_i,j})}-\exp (\beta 'Z_i+V_i\phi (W_i))\right\} h_2(T_{K_i,j})\\&\qquad \qquad \qquad \left. +\,\{N_i(T_{K_i,j})-\Lambda (T_{K_i,j})\exp (\beta 'Z_i+V_i\phi (W_i))\}V_i h_3(W_i)\right] \\&\quad \equiv ~A^{ps}_{n1}(\theta )[\mathbf h _1]+A^{ps}_{n2}(\theta )[h_2]+A^{ps}_{n3}(\theta )[h_3]\\&\quad \equiv ~ P_n(\mathbf h '_1\dot{l}^{ps}_\beta )+P_n(\dot{l}^{ps}_\Lambda [h_2])+P_n(\dot{l}^{ps}_\phi [h_3])\\&\quad \equiv ~ {P}_n \psi _{ps}{(\theta )} [\mathbf h _1,{ h}_2, h_3]. \end{aligned}$$

Correspondingly, we define the limit map \(S^{ps} : \mathcal {U} \longrightarrow l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) as

$$\begin{aligned} S^{ps}(\theta )[\mathbf h _1, {h}_2, h_3] =A^{ps}_1(\theta )[\mathbf h _1]+A^{ps}_2(\theta )[h_2]+A^{ps}_3(\theta )[h_3], \end{aligned}$$

where

$$\begin{aligned}&A^{ps}_1(\theta )[\mathbf h _1]=P\left[ \sum ^{K}_{j=1}\{N(T_{K,j})-\Lambda (T_{K,j})\exp (\beta 'Z+V\phi (W))\} \mathbf h '_1Z\right] ,\\&A^{ps}_2(\theta )[h_2]=P\left[ \sum ^K_{j=1}\left\{ \frac{N(T_{K,j})}{\Lambda (T_{K,j})}-\exp (\beta 'Z+V\phi (W))\right\} h_2(T_{K,j})\right] , \end{aligned}$$

and

$$\begin{aligned} A^{ps}_3(\theta )[h_3]=P\left[ \sum ^K_{j=1}\{N(T_{K,j})-\Lambda (T_{K,j})\exp (\beta 'Z+V\phi (W))\}V h_3(W)\right] . \end{aligned}$$

To derive the asymptotic normality of the estimators \((\hat{\beta }^{ps}_n,\hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n)\), motivated by the proofs of Theorem 3.3.1 of Van der Vaart and Wellner (1996, page 310) and Theorem 2 of Zeng et al. (2005) , we need to verify the following five conditions.

  1. (a1)

    \(\sqrt{n}(S^{ps}_n - S^{ps}) (\hat{\beta }^{ps}_n,\hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n) - \sqrt{n}(S^{ps}_n - S^{ps}) (\beta _0,\Lambda _0, \phi _0) = o_p(1)\).

  2. (a2)

    \(S^{ps}(\beta _0, \Lambda _0, \phi _0) = 0\) and \(S^{ps}_n(\hat{\beta }^{ps}_n, \hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n) = o_p(n^{-1/2}).\)

  3. (a3)

    \(\sqrt{n}(S^{ps}_n - S^{ps}) (\beta _0,\Lambda _0, \phi _0)\) converges in distribution to a tight Gaussian process on \(l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\).

  4. (a4)

    \( S^{ps}(\beta ,\Lambda , \phi )\) is Fr\(\acute{e}\)chet-differentiable at \((\beta _0, \Lambda _0, \phi _0)\) with a continuously invertible derivative \(\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0).\)

  5. (a5)

    \(S^{ps}(\hat{\beta }^{ps}_n,\hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n) -S^{ps}(\beta _0,\Lambda _0, \phi _0)-\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)({\hat{\beta }}^{ps}_n - \beta _0,{\hat{\Lambda }}^{ps}_n - \Lambda _0, {\hat{\phi }}^{ps}_n - \phi _0)=o_p(n^{-1/2}). \)

Condition (a1) holds since

$$\begin{aligned}&\Big \{\psi _{ps} {(\beta , \Lambda , \phi )} [\mathbf h _1,{ h}_2, h_3] - \psi _{ps} {(\beta _0,\Lambda _0, \phi _0)} [\mathbf h _1,{ h}_2, h_3] :\\&\quad \quad \quad \quad d_1((\beta ,\Lambda , \phi ),(\beta _0,\Lambda _0, \phi _0)) < \delta , (\mathbf h _1,{ h}_2, h_3) \in {H_1\times H_2\times \mathcal{F}_r}\Big \} \end{aligned}$$

is a Donsker class for some \(\delta > 0\), and that

$$\begin{aligned} \sup _{(\mathbf h _1,{ h}_2, h_3) \in \mathcal {H}} P \big [\psi _{ps} {( \beta , \Lambda , \phi )} [\mathbf h _1,{ h}_2, h_3] - \psi _{ps} {(\beta _0,\Lambda _0, \phi _0)} [\mathbf h _1,{ h}_2, h_3]\big ]^2 \longrightarrow 0, \end{aligned}$$

as \((\beta ,\Lambda , \phi ) \longrightarrow (\beta _0,\Lambda _0, \phi _0)\) in \(d_1\).

Clearly, \(S^{ps}(\beta _0, \Lambda _0,\phi _0) = 0\). For \(h_3 \in \mathcal {F}_r\), let \(h_{3n}\) be the B-spline function approximation of \(h_3\) with \(||h_3-h_{3n}||_\infty =O(n^{-vr})\) by Corollary 6.21 of Schumaker (1981, page 227). Then we have \(S^{ps}_n(\hat{\beta }^{ps}_n, \hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n)[\mathbf h _1, {h}_2, h_{3n}] = 0\). Thus, for \((\mathbf h _1, {h}_2, h_3) \in {H_1\times H_2\times \mathcal{F}_r}\),

$$\begin{aligned}&\sqrt{n}\{S^{ps}_n(\hat{\beta }^{ps}_n,\hat{\Lambda }^{ps}_n,\hat{\phi }^{ps}_n)[\mathbf h _1, {h}_2, h_{3}]\}\\&\quad =\sqrt{n} P_n\psi _{ps}(\hat{\theta }^{ps}_n)[\mathbf h _1, {h}_2, h_{3}] - \sqrt{n}P_n\psi _{ps}({\hat{\theta }}^{ps}_n)[\mathbf h _1, {h}_2, h_{3n}] \\&\quad =I_{n1}-I_{n2}+I_{n3}+I_{n4} \end{aligned}$$

where

$$\begin{aligned} I_{n1}= & {} \sqrt{n}(P_n-P) \left\{ \psi _{ps}(\hat{\theta }^{ps}_n)[\mathbf h _1, {h}_2, h_{3}] -\psi _{ps}(\theta _0)[\mathbf h _1, {h}_2, h_{3}]\right\} ,\\ I_{n2}= & {} \sqrt{n}(P_n-P) \left\{ \psi _{ps}(\hat{\theta }^{ps}_n)[\mathbf h _1, {h}_2, h_{3n}] -\psi _{ps}(\theta _0)[\mathbf h _1, {h}_2, h_{3n}]\right\} ,\\ I_{n3}= & {} \sqrt{n}P_n\left\{ \psi _{ps}({\theta }_0)[\mathbf h _1, {h}_2, h_{3}] -\psi _{ps}(\theta _0)[\mathbf h _1, {h}_2, h_{3n}]\right\} , \end{aligned}$$

and

$$\begin{aligned} I_{n4}=\sqrt{n}P\left\{ \psi _{ps}({\hat{\theta }}^{ps}_n)[\mathbf h _1, {h}_2, h_{3}] -\psi _{ps}({\hat{\theta }}^{ps}_n)[\mathbf h _1, {h}_2, h_{3n}]\right\} . \end{aligned}$$

From (a1), we have \(I_{n1}=o_p(1)\) and \(I_{n2}=o_p(1)\). Next we need to show \(I_{n3}=o_p(1)\) and \(I_{n4}=o_p(1)\) . Note that

$$\begin{aligned} E(I^2_{n3})=P\left\{ \psi _{ps}({\theta }_0)[\mathbf h _1, {h}_2, h_{3}] -\psi _{ps}(\theta _0)[\mathbf h _1, {h}_2, h_{3n}]\right\} ^2\le c||h_{3n}-h_3||^2_\infty \rightarrow 0, \end{aligned}$$

and

$$\begin{aligned} |I_{n4}|= & {} {\Big |}\sqrt{n} P\left[ \sum ^K_{j=1}\left\{ \Lambda _0(T_{K, j})\exp (Z'\beta _0+V\phi _0(W))\right. \right. \\&\left. \left. -{\hat{\Lambda }}_n^{ps}(T_{K, j})\exp (Z'{\hat{\beta }}^{ps}_n+V{\hat{\phi }}_n^{ps}(W))\right\} V(h_3(W)-h_{3n}(W)\right] {\Big |}\\\le & {} c \sqrt{n} d_1({\hat{\theta }}^{ps}_n, \theta _0) ||h_3-h_{3n}||_\infty \\= & {} O(n^{-\frac{1-v}{3}-vr+\frac{1}{2}}) \end{aligned}$$

by Theorem 3.2. Thus (a2) holds for \(\frac{1}{6r-2}<v<\frac{1}{2}\).

Condition (a3) holds because \({H_1\times H_2\times \mathcal{F}_r}\) is a Donsker class and the functionals \(A^{ps}_{n1}, A^{ps}_{n2}, A^{ps}_{n3}\) are bounded Lipschitz functions with respect to \({H_1\times H_2\times \mathcal{F}_r}\).

For (a4), by the smoothness of \(S^{ps}(\beta ,\Lambda , \phi )\), the Fr\(\acute{e}\)chet differentiability holds and the derivative of \(S^{ps}(\beta ,\Lambda , \phi )\) at \((\beta _0,\Lambda _0, \phi _0)\), denoted by \(\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)\), is a map from the space \(\{( \beta - \beta _0, \Lambda - \Lambda _0,\phi - \phi _0): (\beta ,\Lambda , \phi ) \in \mathcal {U}\}\) to \(l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) and

$$\begin{aligned}&\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1, {h}_2, h_3]\\&\quad =\frac{d}{d\varepsilon }\left\{ A^{ps}_{1}(\theta _0+\varepsilon (\theta -\theta _0))[\mathbf h _1]\right\} \Big |_{\varepsilon =0} +\frac{d}{d\varepsilon }\left\{ A^{ps}_{2}(\theta _0+\varepsilon (\theta -\theta _0))[h_2]\right\} \Big |_{\varepsilon =0}\\&\qquad +\, \frac{d}{d\varepsilon }\left\{ A^{ps}_{3}(\theta _0+\varepsilon (\theta -\theta _0))[h_3]\right\} \Big |_{\varepsilon =0}\\&\quad =-P\sum ^K_{j=1}\exp (\beta _0'Z+V\phi _0(W)) \mathbf{h}_1'Z \left[ \left\{ \Lambda (T_{K, j})-\Lambda _0(T_{K,j})\right\} \right. \\&\left. \qquad +\, \Lambda _0(T_{K,j}) \left\{ (\beta -\beta _0)'Z+V(\phi (W)-\phi _0(W)) \right\} \right] \\&\qquad -P\sum ^K_{j=1}\exp (\beta _0'Z+V\phi _0(W)) h_2(T_{K, j}) \left[ \left\{ \frac{\Lambda (T_{K, j})- \Lambda _0(T_{K,j})}{\Lambda _0(T_{K, j})}\right. \right. \\&\left. \qquad + \left\{ (\beta -\beta _0)'Z+V(\phi (W)-\phi _0(W)) \right\} \right] \\&\qquad -P\sum ^K_{j=1}\exp (\beta _0'Z+V\phi _0(W)) Vh_3(W) \left[ \left\{ \Lambda (T_{K, j})-\Lambda _0(T_{K,j})\right\} \right. \\&\left. \qquad +\Lambda _0(T_{K,j}) \left\{ (\beta -\beta _0)'Z+V(\phi (W)-\phi _0(W)) \right\} \right] . \end{aligned}$$

Thus, we have

$$\begin{aligned}&\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0) [\mathbf h _1, {h}_2, h_3]\nonumber \\&\quad =(\beta -\beta _0)'Q^{ps}_1(\mathbf{h}_1,h_2,h_3)+ \int (\Lambda (t)-\Lambda _0(t)) dQ^{ps}_2(\mathbf{h}_1, h_2, h_3)(t) \nonumber \\&\qquad +\int (\phi (w)-\phi _0(w))dQ^{ps}_3(\mathbf{h}_1, h_2, h_3)(w) \end{aligned}$$
(7.4)

where

$$\begin{aligned} Q^{ps}_1(\mathbf h _1,h_2, h_3)= & {} -E\left[ Z\exp \{\beta _0'Z+V\phi _0(W)\} \sum ^K_{j=1} \left\{ \Lambda _0(T_{K,j})\mathbf h _1'Z+h_2(T_{K,j})\right. \right. \\&\left. \left. +\,\Lambda _0(T_{K,j})V h_3(W)\right\} \right] ,\\ dQ^{ps}_2(\mathbf h _1,h_2, h_3)(t)= & {} -E\left[ \exp \{\beta _0'Z+V\phi _0(W)\} \sum ^K_{j=1}\frac{1}{\Lambda _0(t)} \left\{ \Lambda _0(t)\mathbf h _1'Z+h_2(t)\right. \right. \\&\left. \left. +\,\Lambda _0(t)V h_3(W)\right\} dP(T_{K,j}\le t|K, Y)\right] , \end{aligned}$$

and

$$\begin{aligned}&dQ^{ps}_3(\mathbf h _1,h_2, h_3)(w)\\&\quad =-E\left[ V\exp \{\beta _0'Z+V\phi _0(w)\} \sum ^K_{j=1}\left\{ \Lambda _0(T_{K,j})\mathbf h _1'Z+h_2(T_{K,j})\right. \right. \\&\qquad \left. \left. +\, \Lambda _0(T_{K,j})V h_3(w)\right\} |W=w\right] dF_W(w) \end{aligned}$$

where \(F_W\) denotes the cumulative distribution of W.

Next, we show that \(Q^{ps}=(Q^{ps}_1, Q^{ps}_2, Q^{ps}_3)\) is one-to-one, that is, for \((\mathbf h _1,h_2, h_3) \in {H_1\times H_2\times \mathcal{F}_r}\), if \(Q^{ps}(\mathbf h _1,h_2, h_3) = 0\), then \(\mathbf h _1=\mathbf{0},h_2=0, h_3=0\).

Suppose that \(Q^{ps}(\mathbf h _1,h_2, h_3) = 0\). Then \(\dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1, {h}_2, h_3]=0\) for any \((\beta , \Lambda , \phi )\) in the neighborhood \(\mathcal U\). In particular, we take \(\beta =\beta _0+\epsilon \mathbf{h}_1, \Lambda =\Lambda _0+\epsilon h_2, \phi =\phi _0+\epsilon h_3\) for a small constant \(\epsilon \). Thus we have

$$\begin{aligned} 0= & {} \dot{S}^{ps}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1,h_2, h_3]\\= & {} -\,\epsilon E\left[ \exp \{\beta _0'Z+V\phi _0(W)\}\sum ^K_{j=1}{\Lambda _0(T_{K,j})} \left\{ \mathbf h _1'Z+Vh_3(W)+\frac{h_2(T_{K,j})}{\Lambda _0(T_{K,j})}\right\} ^2\right] , \end{aligned}$$

which yields

$$\begin{aligned} \mathbf h _1'Z+Vh_3(W)+\frac{h_2(T_{K,j})}{\Lambda _0(T_{K,j})}=0, \; j=1, \ldots , K, \; \; a.s. \end{aligned}$$

and so \(\mathbf h _1=\mathbf{0},h_2=0, h_3=0\) by C7.

Next we show that (a5) holds. Write

$$\begin{aligned}&S^{ps}({\hat{\theta }}^{ps}_n)[\mathbf{h}_1, h_2, h_3]-S^{ps}(\theta _0)[\mathbf{h}_1, h_2, h_3]\\&\qquad -\,\dot{S}(\beta _0,\Lambda _0, \phi _0)({\hat{\beta }}^{ps}_n - \beta _0,{\hat{\Lambda }}^{ps}_n - \Lambda _0, {\hat{\phi }}^{ps}_n - \phi _0)[\mathbf h _1,h_2, h_3]\\&\quad =B_{n1}+B_{n2}+B_{n3} \end{aligned}$$

where

$$\begin{aligned} B_{n1}= & {} A^{ps}_1({\hat{\theta }}^{ps}_n)[\mathbf h _1]-\frac{d}{d\varepsilon }\left\{ A^{ps}_{1}(\theta _0+\varepsilon ({\hat{\theta }}^{ps}_n-\theta _0))[\mathbf h _1]\right\} \Big |_{\varepsilon =0} \, ,\\ B_{n2}= & {} A^{ps}_2({\hat{\theta }}^{ps}_n)[h_2] -\frac{d}{d\varepsilon }\left\{ A^{ps}_{2}(\theta _0+\varepsilon ({\hat{\theta }}^{ps}_n-\theta _0))[h_2]\right\} \Big |_{\varepsilon =0} \, , \end{aligned}$$

and

$$\begin{aligned} B_{n3}=A^{ps}_3({\hat{\theta }}^{ps}_n)[h_3]-\frac{d}{d\varepsilon }\left\{ A^{ps}_{3}(\theta _0+\varepsilon ({\hat{\theta }}^{ps}_n-\theta _0))[h_3]\right\} \Big |_{\varepsilon =0} \end{aligned}$$

It is easy to show that \(B_{n1}=O_p(d_1^{2}({\hat{\theta }}^{ps}_n, \theta _0))\), \(B_{n2}=O_p(d_1^{2}({\hat{\theta }}^{ps}_n, \theta _0))\), and \(B_{n3}=O_p(d_1^{2}({\hat{\theta }}^{ps}_n, \theta _0))\). Thus, by Theorem  3.2, (a5) holds for \(0<v<1/4\).

It follows from (7.4), (a1), (a2) and (a5) that

$$\begin{aligned}&\sqrt{n}({\hat{\beta }}^{ps}_n-\beta _0)'Q^{ps}_1(\mathbf{h}_1,h_2,h_3) +\sqrt{n}\int \{{\hat{\Lambda }}^{ps}_n(t)-\Lambda _0(t)\} dQ^{ps}_2(\mathbf{h}_1, h_2, h_3)(t) \\&\qquad +\,\sqrt{n}\int \{{\hat{\phi }}^{ps}_n(w)-\phi _0(w)\}d Q^{ps}_3(\mathbf{h}_1, h_2, h_3) (w) \\&\quad = - \sqrt{n} ( S^{ps}_n - S^{ps}) (\beta _0,\Lambda _0, \phi _0) [\mathbf h _1, {h}_2, h_3] + o_p(1), \end{aligned}$$

uniformly in \(\mathbf h _1\), \( {h}_2\) and \(h_3\).

For each \((\mathbf h _1, {h}_2, h_3)\in {H_1\times H_2\times \mathcal{F}_r}\), since \(Q^{ps}\) is invertible, there exists \((\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3)\in {H_1\times H_2\times \mathcal{F}_r}\) such that

$$\begin{aligned} Q^{ps}_1(\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3)=\mathbf h _1, Q^{ps}_2(\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3)={ h}_2, Q^{ps}_3(\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3) = h_3. \end{aligned}$$

Therefore, we have

$$\begin{aligned}&\mathbf h '_1\sqrt{n}(\hat{\beta }^{ps}_n -\beta _0) +\sqrt{n} \int \{{\hat{\Lambda }}^{ps}_n(t)-\Lambda _0(t)\}d h_2(t) \\&\qquad +\, \sqrt{n} \int \{{\hat{\phi }}^{ps}_n(w)-\phi _0(w)\} d h_3(w) \\&\quad = - \sqrt{n} ( S^{ps}_n - S^{ps}) (\beta _0,\Lambda _0, \phi _0) [\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3] + o_p(1)\\&\quad \rightarrow _d N(0, \sigma ^2_{ps}), \end{aligned}$$

where

$$\begin{aligned} \sigma ^2_{ps}=E\{\psi ^2_{ps}{(\beta _0, \Lambda _0, \phi _0)}[\mathbf h ^{ps}_1, {h}^{ps}_2, h^{ps}_3]\}. \end{aligned}$$
(7.5)

To prove part (ii), we define a sequence of maps \(S_n\) mapping a neighborhood of \((\beta _0,\Lambda _0, \phi _0)\), \(\mathcal {U}\), in the parameter space for \((\beta ,\Lambda , \phi )\) into \(l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) as:

$$\begin{aligned} S_n(\theta )[\mathbf h _1,{h}_2, h_3] =n^{-1} \frac{d}{d\varepsilon } l_n( \beta + \varepsilon \mathbf h _1,\Lambda + \varepsilon {h}_2 , \phi + \varepsilon h_3)\Big |_{\varepsilon =0}. \end{aligned}$$

Write \(\Delta N_i(T_{K_i,j})=N_i(T_{K_i,j})-N_i(T_{K_i,j-1})\), \(\Delta \Lambda (T_{K_i,j})=\Lambda (T_{K_i,j})-\Lambda (T_{K_i,j-1})\), and \(\Delta h(T_{K_i,j})=h(T_{K_i,j})-h(T_{K_i,j-1})\).

Then, we have

$$\begin{aligned}&S_n(\theta )[\mathbf h _1,{h}_2, h_3]\\&\quad =n^{-1} \sum ^n_{i=1}\sum ^{K_i}_{j=1}\left[ \{\Delta N_i(T_{K_i,j})-\Delta \Lambda (T_{K_i,j})\exp (\beta 'Z_i+V_i\phi (W_i))\}{} \mathbf h '_1Z_i\right. \\&\qquad +\left\{ \frac{\Delta N_i(T_{K_i,j})}{\Delta \Lambda (T_{K_i,j})}-\exp (\beta 'Z_i+V_i\phi (W_i))\right\} \Delta h_2(T_{K_i,j})\\&\qquad \left. +\, \{\Delta N_i(T_{K_i,j})-\Delta \Lambda (T_{K_i,j})\exp (\beta 'Z_i+V_i\phi (W_i))\}V_i h_3(W_i)\right] \\&\quad \equiv ~A_{n1}(\theta )[\mathbf h _1]+A_{n2}(\theta )[h_2]+A_{n3}(\theta )[h_3]\\&\quad \equiv ~ P_n(\mathbf h '_1\dot{l}_\beta )+P_n(\dot{l}_\Lambda [h_2])+P_n(\dot{l}_\phi [h_3])\\&\quad \equiv ~ {P}_n \psi {(\theta )} [\mathbf h _1,{ h}_2, h_3]. \end{aligned}$$

Correspondingly, we define the limit map \(S : \mathcal {U} \longrightarrow l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) as

$$\begin{aligned} S(\theta )[\mathbf h _1, {h}_2, h_3] =A_1(\theta )[\mathbf h _1]+A_2(\theta )[h_2]+A_3(\theta )[h_3], \end{aligned}$$

where

$$\begin{aligned}&A_1(\theta )[\mathbf h _1]=E\left[ \sum ^{K}_{j=1}\{\Delta N(T_{K,j})-\Delta \Lambda (T_{K,j})\exp (\beta 'Z+V\phi (W))\} \mathbf h '_1Z\right] ,\\&A_2(\theta )[h_2]=E\left[ \sum ^K_{j=1}\left\{ \frac{\Delta N(T_{K,j})}{\Delta \Lambda (T_{K,j})}-\exp (\beta 'Z+V\phi (W))\right\} \Delta h_2(T_{K,j})\right] , \end{aligned}$$

and

$$\begin{aligned} A_3(\theta )[h_3]=E\left[ \sum ^K_{j=1}\{\Delta N(T_{K,j})-\Delta \Lambda (T_{K,j})\exp (\beta 'Z+V\phi (W))\}V h_3(W)\right] . \end{aligned}$$

Furthermore, the derivative of \(S(\beta ,\Lambda , \phi )\) at \((\beta _0,\Lambda _0, \phi _0)\), denoted by \(\dot{S}(\beta _0,\Lambda _0, \phi _0)\), is a map from the space \(\{( \beta - \beta _0, \Lambda - \Lambda _0,\phi - \phi _0): (\beta ,\Lambda , \phi ) \in \mathcal {U}\}\) to \(l^{\infty }({H_1\times H_2\times \mathcal{F}_r})\) and

$$\begin{aligned}&\dot{S}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1, {h}_2, h_3]\\&\quad =(\beta -\beta _0)'Q_1(\mathbf{h}_1,h_2,h_3)+\int \{\Lambda (t)- \Lambda _0(t) \} dQ_2(\mathbf{h}_1, h_2, h_3)(t) \\&\qquad +\int \{\phi (w)-\phi _0(w)\}dQ_3(\mathbf{h}_1, h_2, h_3)(w) \end{aligned}$$

where

$$\begin{aligned}&Q_1(\mathbf h _1,h_2, h_3) \\&\quad =-E\left[ Z\exp \{\beta _0'Z+V\phi _0(W)\} \right. \\&\left. \qquad \times \sum ^K_{j=1} \left\{ \Delta \Lambda _0(T_{K,j})\mathbf h _1'Z+\Delta h_2(T_{K,j}) +\Delta \Lambda _0(T_{K,j})V h_3(W)\right\} \right] ,\\&dQ_2(\mathbf h _1,h_2, h_3)(t)\\&\quad =- E\left[ \exp \{\beta _0'Z+V\phi _0(W)\} \right. \\&\qquad \times \sum ^K_{j=1} \left\{ \left( \mathbf h _1'Z+\frac{h_2(t)-h_2(T_{K, j-1})}{\Lambda _0(t)-\Lambda _0(T_{K, j-1})}\!+\! V h_3(W)\right) dP(T_{K, j}\le t|K, T_{K,j-1},Y)\right. \\&\qquad \left. \left. -\left( \mathbf h _1'Z+\frac{h_2(T_{K, j})-h_2(t)}{\Lambda _0(T_{K, j})-\Lambda _0(t)}+ V h_3(W)\right) dP(T_{K, j-1}\le t|K, T_{K,j},Y)\right\} \right] , \end{aligned}$$

and

$$\begin{aligned}&dQ_3(\mathbf h _1,h_2, h_3)(w) \\&\quad =-E\left[ V\exp \{\beta _0'Z+V\phi _0(w)\} \right. \\&\qquad \times \left. \sum ^K_{j=1}\left\{ \Delta \Lambda _0(T_{K,j})\mathbf h _1'Z{+}\Delta h_2(T_{K,j})\!+\!\Delta \Lambda _0(T_{K,j})V h_3(w)\right\} |W=w\right] dF_W(w). \end{aligned}$$

Next, we show that \(Q=(Q_1, Q_2, Q_3)\) is one-to-one, that is, for \((\mathbf h _1,h_2, h_3) \in {H_1\times H_2\times \mathcal{F}_r}\), if \(Q(\mathbf h _1,h_2, h_3) = 0\), then \(\mathbf h _1=\mathbf{0},h_2=0, h_3=0\)

Suppose that \(Q(\mathbf h _1,h_2, h_3) = 0\). Then \(\dot{S}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1, {h}_2, h_3]=0\) for any \((\beta , \Lambda , \phi )\) in the neighborhood \(\mathcal U\). In particular, we take \(\beta =\beta _0+\epsilon \mathbf{h}_1, \Lambda =\Lambda _0+\epsilon h_2, \phi =\phi _0+\epsilon h_3\) for a small constant \(\epsilon \). Thus we have

$$\begin{aligned} 0= & {} \dot{S}(\beta _0,\Lambda _0, \phi _0)(\beta - \beta _0,\Lambda - \Lambda _0, \phi - \phi _0)[\mathbf h _1,h_2, h_3]\\= & {} -\,\epsilon E\left[ \exp \{\beta _0'Z\!+\!V\phi _0(W)\}\sum ^K_{j=1}{\Delta \Lambda _0(T_{K,j})} \left\{ \mathbf h _1'Z\!+\!Vh_3(W)+\frac{\Delta h_2(T_{K,j})}{\Delta \Lambda _0(T_{K,j})}\right\} ^2\right] , \end{aligned}$$

which yields

$$\begin{aligned} \mathbf h _1'Z+Vh_3(W)+\frac{\Delta h_2(T_{K,j})}{\Delta \Lambda _0(T_{K,j})}=0, \; j=1, \ldots , K, \; \; a.s. \end{aligned}$$

and so \(\mathbf h _1=\mathbf{0},h_2=0, h_3=0\) by C7.

Similarly, we can show that \(S(\beta _0, \Lambda _0, \phi _0) = 0\), \(S_n(\hat{\beta }_n, \hat{\Lambda }_n,\hat{\phi }_n) = o_p(n^{-1/2})\), and

$$\begin{aligned}&S({\hat{\theta }}_n)[\mathbf{h}_1, h_2, h_3]-S(\theta _0)[\mathbf{h}_1, h_2, h_3]\\&\quad =\dot{S}(\beta _0,\Lambda _0, \phi _0)({\hat{\beta }}_n - \beta _0,{\hat{\Lambda }}_n - \Lambda _0, {\hat{\phi }}_n - \phi _0)[\mathbf h _1,h_2, h_3] +O_p(d_2^2({\hat{\theta }}_n,\theta _0))\\&\quad =\dot{S}(\beta _0,\Lambda _0, \phi _0)({\hat{\beta }}_n - \beta _0,{\hat{\Lambda }}_n - \Lambda _0, {\hat{\phi }}_n - \phi _0)[\mathbf h _1,h_2, h_3]+o_p(n^{-1/2}). \end{aligned}$$

for \(0<v<1/4\). Thus it follows that

$$\begin{aligned}&\sqrt{n}({\hat{\beta }}_n-\beta _0)'Q_1(\mathbf{h}_1,h_2,h_3) +\sqrt{n}\int \{{\hat{\Lambda }}_n(t)- \Lambda _0(t)\} dQ_2(\mathbf{h}_1, h_2, h_3)(t) \\&\qquad +\,\sqrt{n}\int \{{\hat{\phi }}_n(w)-\phi _0(w)\}d Q_3(\mathbf{h}_1, h_2, h_3) (w) \\&\quad = -\, \sqrt{n} ( S_n - S) (\beta _0,\Lambda _0, \phi _0) [\mathbf h _1, {h}_2, h_3] + o_p(1), \end{aligned}$$

uniformly in \(\mathbf h _1\), \( {h}_2\) and \(h_3\).

For each \((\mathbf h _1, h_2, h_3)\in (\mathbf h _1,h_2, h_3)\), since Q is invertible, there exists \((\mathbf h ^{*}_1, \mathbf h ^*_2, h^{*}_3)\in (\mathbf h _1,h_2, h_3)\) such that

$$\begin{aligned} Q^{*}_1(\mathbf h ^{*}_1, {h}^{*}_2, h^{*}_3) =\mathbf h _1, Q^{*}_2(\mathbf h ^{*}_1, {h}^{*}_2, h^{*}_3) =0, Q^{*}_3(\mathbf h ^{*}_1, {h}^{*}_2, h^{*}_3) = h_3. \end{aligned}$$

Thus, we have

$$\begin{aligned}&\mathbf h '_1\sqrt{n}(\hat{\beta }_n -\beta _0) +\sqrt{n} \int \{{\hat{\Lambda }}_n(t)-\Lambda _0(t)\}d h_2(t) \\&\quad +\, \sqrt{n} \int \{{\hat{\phi }}_n(w)-\phi _0(w) \}d h_3(w) \\&\quad = -\, \sqrt{n} ( S_n - S) (\beta _0,\Lambda _0, \phi _0) [\mathbf h ^{*}_1, {h}^{*}_2, h^{*}_3] + o_p(1)\\&\quad \rightarrow _d N(0, \sigma ^2), \end{aligned}$$

where

$$\begin{aligned} \sigma ^2=E\{\psi ^2{(\beta _0, \Lambda _0, \phi _0)}[\mathbf h ^{*}_1, {h}^{*}_2, h^{*}_3]\}. \end{aligned}$$
(7.6)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

He, X., Feng, X., Tong, X. et al. Semiparametric partially linear varying coefficient models with panel count data. Lifetime Data Anal 23, 439–466 (2017). https://doi.org/10.1007/s10985-016-9368-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10985-016-9368-x

Keywords

Navigation