The Long Term Fréchet distribution: Estimation, Properties and its

Application

Pedro Luiz Ramos, Diego Nascimento and Francisco Louzada

Institute of Mathematical Science and Computing, University of São Paulo,

São Carlos, Brazil

Abstract

In this paper a new long-term survival distribution is proposed. The so called long term
Fréchet distribution allows us to fit data where a part of the population is not susceptible
to the event of interest. This model may be used, for example, in clinical studies where a
portion of the population can be cured during a treatment. It is shown an account of
mathematical properties of the new distribution such as its moments and survival
properties. As well is presented the maximum likelihood estimators (MLEs) for the
parameters. A numerical simulation is carried out in order to verify the performance of
the MLEs. Finally, an important application related to the leukemia free-survival times
for transplant patients are discussed to illustrates our proposed distribution

Keywords: Fréchet distribution, Long-term survival distribution, Survival model.

1. Introduction

Extreme value models play an important role in statistic. The generalized extreme value
(GEV) distribution (Jenkinson, 1955) and its sub-models are widely used in application
involving extreme events. These sub-models are the well known Weibull, Fréchet and
Gumbel distributions. The Fréchet distribution can be seen as the inverse Weibull
distribution which gives a probability density function (PDF) such as
𝛼𝛼 𝑡𝑡 −(𝛼𝛼+1) −� 𝑡𝑡 �−𝛼𝛼
𝑓𝑓(𝑡𝑡, 𝜆𝜆, 𝛼𝛼) = � � 𝑒𝑒 𝜆𝜆 . (1)
𝜆𝜆 𝜆𝜆
The survival function is given by
𝑡𝑡 −𝛼𝛼
𝑆𝑆(𝑡𝑡, 𝜆𝜆, 𝛼𝛼) = 1 − 𝑒𝑒 −�𝜆𝜆 � . (2)

1
Although the GEV distribution is the most used generalization of the Fréchet model,
other distributions has been proposed in the literature. De Gusmão (2012) proposed a
three parameter generalized inverse Weibull distribution in which includes the Fréchet
distribution. Krishna et al. (2013) proposed the Marshall-Olkin Fréchet distribution.
Barreto-Souza et al. (2011) discussed some results for beta Fréchet distribution.
However, in survival studies habitually the researches may consider a portion of the
population as cured during a given treatment, this type of distribution is called long-
term (LT) survival models.

In this study, a long-term survival novel proposing a mixture model introduced by

Berkson and Gage (1952), hereafter we shall call it the long-term Fréchet distribution or
simplistically the LF distribution. Some mathematical properties about the LF
distribution were provided such as moments, survival properties and hazard function.
The maximum likelihood estimators of the parameters and its asymptotic properties are
discussed likewise. Similar studies were presented by Roman et al. (2012) for the
geometric exponential distribution and by Louzada and Ramos (2017) for the weighted
Lindley distribution. It was performed a numerical simulation towards to examine the
performance of the MLEs. Finally, our proposed methodology is illustrated in a real
data set related to the leukemia free-survival times (in years) for the 50 autologous
transplant patients.

The paper is organized as follows. Section 2 presents the long term Fréchet distribution
and its mathematical properties. Section 3 discusses the parameter estimation under the
maximum likelihood approach. Section 4 presents a simulation study under different
values of the parameters and different levels of censorship. The proposed methodology
is also fully illustrated in a real data set. Lastly, Section 6 summarizes the founds in this
study and its potential contribution.

2. Long Term Fréchet distribution

Long-term survivors are an important feature to incorporate in the modeling process,

since a portion of the population may no longer be eligible to the event of interest
(according to Maller and Zhou, 1995; or Perdona and Louzada, 2011). Hence the
population can be segregate as a not eligible to the event of interest with probability 𝑝𝑝

2
and as eligible (in risk) to the event of interest with probability (1 − 𝑝𝑝). The long-term
survivor is expressed as

𝑆𝑆(𝑡𝑡; 𝑝𝑝, 𝜽𝜽) = 𝑝𝑝 + (1 − 𝑝𝑝)𝑆𝑆0 (𝑡𝑡; 𝜽𝜽), (3)

where 𝑝𝑝 ∈ (0,1) and 𝑆𝑆0 (𝑡𝑡; 𝜽𝜽) is the survival function related to the eligible group. The
obtained survival function (not conditional) is improper and its limit corresponds to the
individual proportion cure. From the survival function one can easily derive the PDF
(improper) given by

𝜕𝜕
𝑓𝑓(𝑡𝑡; 𝑝𝑝, 𝜽𝜽) = − 𝑆𝑆(𝑡𝑡; 𝑝𝑝, 𝜽𝜽) = (1 − 𝜋𝜋)𝑓𝑓0 (𝑡𝑡; 𝜽𝜽), (4)
𝜕𝜕𝜕𝜕

where 𝑓𝑓0 (𝑡𝑡; 𝜽𝜽) is the PDF related to the susceptible group.

Considering that 𝑓𝑓0 (𝑡𝑡; 𝜽𝜽) follows a Fréchet distribution, then the PDF of the Long Term
Fréchet (LF) distribution is given by

𝑡𝑡 −𝛼𝛼
𝛼𝛼(1−𝑝𝑝) 𝑡𝑡 −(𝛼𝛼+1) −�𝜆𝜆 �
𝑓𝑓(𝑡𝑡; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = � � 𝑒𝑒 , (5)
𝜆𝜆 𝜆𝜆

where 𝜆𝜆 > 0, 𝛼𝛼 > 0 and 𝑝𝑝 ∈ (0,1). The cumulative distribution function is given by

𝑡𝑡 −𝛼𝛼
𝐹𝐹(𝑡𝑡; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = (1 − 𝑝𝑝)𝑒𝑒 −�𝜆𝜆 � . (6)

In this case, the LF has the quantile function in closed-form and is given by

1
−
(1 − 𝑝𝑝) 𝛼𝛼
𝑡𝑡𝑢𝑢 = 𝜆𝜆 log � � , (7)
𝑢𝑢
where 0 ≤ 𝑢𝑢 < 1. The 𝑟𝑟-th moments of T about the origin is

𝑟𝑟
𝐸𝐸(𝑇𝑇 𝑟𝑟 ; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = (1 − 𝑝𝑝)λr Γ �1 − � , 𝛼𝛼 > 𝑟𝑟, (8)
𝛼𝛼
∞
for𝑟𝑟 ∈ ℕ and Γ(𝑥𝑥) = ∫0 𝑒𝑒 −𝑦𝑦 𝑦𝑦 𝑥𝑥−1 𝑑𝑑𝑑𝑑is called gamma function. Along with some
algebraic manipulation the mean and variance of the LF distribution are given,
respectively, by

1
𝐸𝐸(𝑇𝑇; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = (1 − 𝑝𝑝)λΓ �1 − � , 𝛼𝛼 > 1
𝛼𝛼
and

3
 2
2 1 2
𝑉𝑉(𝑇𝑇; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = (1 − 𝑝𝑝)𝜆𝜆 �Γ �1 − � − (1 − 𝑝𝑝)Γ �1 − � � , 𝛼𝛼 > 2.
𝛼𝛼 𝛼𝛼

The survival and hazard functions of 𝐿𝐿𝐿𝐿(𝜆𝜆, 𝛼𝛼, 𝑝𝑝) distribution is given by

𝑡𝑡 −𝛼𝛼
𝑆𝑆(𝑡𝑡; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = 𝑝𝑝 + (1 − 𝑝𝑝) �1 − 𝑒𝑒 −�𝜆𝜆 � � (9)

and

𝑡𝑡 −𝛼𝛼
𝛼𝛼 𝑡𝑡 −(𝛼𝛼 +1) −� �
� � 𝑒𝑒 𝜆𝜆
𝜆𝜆 𝜆𝜆
ℎ(𝑡𝑡; 𝜆𝜆, 𝛼𝛼, 𝑝𝑝) = 𝑡𝑡 −𝛼𝛼
. (10)
𝑝𝑝+(1−𝑝𝑝)�1−exp �−� � ��
𝜆𝜆

Figure 1 shows some cases about the PDF and the survival function shapes applied to
LF distribution.

Figure 2. Left panel: Probability density function of the LF distribution. Right panel:
Survival function of the LF distribution

3. Parameter Estimation

For each failure time related to the i-th individual, it may not be perceived or subject by
the right censoring. Furthermore, the random censoring times 𝐶𝐶𝑖𝑖 s are independent of 𝑇𝑇𝑖𝑖 s
(non-censored time) and their distribution does not depend on the parameters. In a
scenario of a 𝑛𝑛 sample of size, the data set will be describe by 𝒟𝒟 = (𝑡𝑡𝑖𝑖 , 𝛿𝛿𝑖𝑖 ), where

4
𝑡𝑡𝑖𝑖 = min(𝑇𝑇𝑖𝑖 , 𝐶𝐶𝑖𝑖 ) and 𝛿𝛿𝑖𝑖 = 𝐼𝐼(𝑇𝑇𝑖𝑖 ≤ 𝐶𝐶𝑖𝑖 ). This general random censoring scheme has as
special case type I and II censoring mechanism. The likelihood function is given by

𝑛𝑛

𝐿𝐿(𝜽𝜽; 𝓓𝓓) = � 𝑓𝑓(𝑡𝑡𝑖𝑖 ; 𝜽𝜽)𝛿𝛿 𝑖𝑖 𝑆𝑆(𝑡𝑡𝑖𝑖 ; 𝜽𝜽)1−𝛿𝛿 𝑖𝑖 .

𝑖𝑖=1

Let 𝑇𝑇1 , … , 𝑇𝑇𝑛𝑛 be a random sample of LF distribution, the likelihood function considering
data with random censoring is given by

𝑛𝑛 𝑛𝑛
𝛼𝛼 𝑑𝑑 (1 − 𝑝𝑝)𝑑𝑑 −𝛿𝛿 𝑖𝑖 (𝛼𝛼+1) 𝑡𝑡𝑖𝑖 −𝛼𝛼
𝐿𝐿(𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = � 𝑡𝑡𝑖𝑖 exp �− � 𝛿𝛿𝑖𝑖 � � �
𝜆𝜆−𝑑𝑑𝑑𝑑 𝜆𝜆
𝑖𝑖=1 𝑖𝑖=1
1−𝛿𝛿 𝑖𝑖
𝑡𝑡 −𝛼𝛼
−�𝜆𝜆 �
× �𝑝𝑝 + (1 − 𝑝𝑝) �1 − 𝑒𝑒 �� ,

where 𝑑𝑑 = ∑𝑛𝑛𝑖𝑖=1 𝛿𝛿𝑖𝑖 . The log-likelihood function is given as

𝑛𝑛

𝑙𝑙(𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = 𝑑𝑑 log(𝛼𝛼) + 𝑑𝑑 log(1 − 𝑝𝑝) + 𝑑𝑑𝑑𝑑 log 𝜆𝜆 − (𝛼𝛼 + 1) � 𝛿𝛿𝑖𝑖 log 𝑡𝑡𝑖𝑖
𝑖𝑖=1
𝑛𝑛 𝑛𝑛
(11)
𝛼𝛼
𝜆𝜆 𝑡𝑡 −𝛼𝛼
−�𝜆𝜆 �
− � 𝛿𝛿𝑖𝑖 � � + �(1 − 𝛿𝛿𝑖𝑖 ) log �𝑝𝑝 + (1 − 𝑝𝑝) �1 − 𝑒𝑒 ��.
𝑡𝑡𝑖𝑖
𝑖𝑖=1 𝑖𝑖=1

The maximum likelihood estimators (MLEs) are widely explored as statistical

inferential methodology due its many desirable properties, in which includes
consistency, asymptotic efficiency and invariance. The MLEs are obtained from the
maximization of the log-likelihood function (11). Before we derive the MLEs of the LF,
let us define the following function

𝑛𝑛
log 𝑆𝑆(𝑡𝑡𝑖𝑖 ; 𝛉𝛉)
𝜂𝜂𝑗𝑗 (𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = �(1 − 𝛿𝛿𝑖𝑖 ) , 𝑗𝑗 = 1,2,3.
𝜕𝜕𝜃𝜃𝑗𝑗
𝑖𝑖=1

Then, the likelihood equations are given by

𝑑𝑑𝑑𝑑
− 𝛼𝛼𝜆𝜆𝛼𝛼−1 ∑𝑛𝑛𝑖𝑖=1 𝛿𝛿𝑖𝑖 𝑡𝑡𝑖𝑖−𝛼𝛼 + 𝜂𝜂1 (𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = 0,
𝜆𝜆

𝑛𝑛 𝑛𝑛
𝑑𝑑 𝜆𝜆 𝛼𝛼 𝜆𝜆
+ 𝑑𝑑 log 𝜆𝜆 − � 𝛿𝛿𝑖𝑖 � � log � � − � 𝛿𝛿𝑖𝑖 log 𝑡𝑡𝑖𝑖 + 𝜂𝜂2 (𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = 0
𝛼𝛼 𝑡𝑡𝑖𝑖 𝑡𝑡𝑖𝑖
𝑖𝑖=1 𝑖𝑖=1

and

5
 𝑑𝑑
+ 𝜂𝜂3 (𝜆𝜆, 𝛼𝛼, 𝑝𝑝; 𝓓𝓓) = 0.
𝑝𝑝−1

The maximization of the log-likelihood function can be performed directly by using

existing statistical packages. Further information about the numerical procedures will be
discussed in the next section.

According to Migon et al. (2014), under mild conditions the obtained estimators are
consistent and efficient with an asymptotically normal joint distribution given by

�𝜆𝜆̂, 𝛼𝛼�, 𝑝𝑝̂ �~𝑁𝑁3 �(𝜆𝜆, 𝛼𝛼, 𝑝𝑝), 𝐻𝐻 −1 (𝜆𝜆, 𝛼𝛼, 𝑝𝑝)�,

where 𝑯𝑯(𝝀𝝀, 𝜶𝜶, 𝒑𝒑) is the 3×3 observed Fisher information matrix and 𝑯𝑯𝒊𝒊𝒊𝒊 (𝝀𝝀, 𝜶𝜶, 𝒑𝒑) is the
Fisher information given by
𝜕𝜕
𝐻𝐻𝑖𝑖𝑖𝑖 (𝜽𝜽) = − 𝑙𝑙(𝜽𝜽; 𝓓𝓓), 𝑖𝑖, 𝑗𝑗 = 1,2,3.
𝜕𝜕𝜃𝜃𝑖𝑖 𝜕𝜕𝜃𝜃𝑗𝑗

Note that, the observed Fisher information matrix was used since it is not possible to
compute the expected Fisher information matrix due its lack of closed form expression.
For large samples, confidence intervals approximation can be constructed for the
individual parameters 𝜽𝜽𝒊𝒊 i=1,2,3, assuming a confidence coefficient 𝟏𝟏𝟏𝟏𝟏𝟏(𝟏𝟏 − 𝜸𝜸)% the
marginal distributions are given by

𝜃𝜃�𝑖𝑖 ~𝑁𝑁 �𝜃𝜃𝑖𝑖 , 𝐻𝐻𝑖𝑖𝑖𝑖−1 (𝜽𝜽)� , 𝑖𝑖 = 1,2,3.

4. Simulation Study

The maximum likelihood method efficiency was analyzed through a simulation study
on the LF distribution. This procedure was conducted by computing the mean relative
errors (MRE) and the mean square errors (MSE) given by
𝑁𝑁 𝑁𝑁
1 𝜃𝜃�𝑖𝑖,𝑗𝑗 1 2
𝑀𝑀𝑀𝑀𝐸𝐸𝑖𝑖 = � , 𝑀𝑀𝑀𝑀𝐸𝐸𝑖𝑖 = ��𝜃𝜃�𝑖𝑖,𝑗𝑗 − 𝜃𝜃𝑖𝑖 � , for 𝑖𝑖 = 1,2,3.
𝑁𝑁 𝜃𝜃𝑖𝑖 𝑁𝑁
𝑗𝑗 =1 𝑗𝑗 =1

as N is the number of estimates obtained through the MLE approach. The 95%
coverage probabilities of the asymptotic confidence intervals were also evaluated. The
adopted approach prioritize that the expected MLEs returns the MREs closer to one
with smaller MSEs. Additionally, by considering a 95% confidence level, the interval

6
covers the true values of θ closer to 95%.Considering scenarios with sample sizes
n=(10,25,50,100, 200) and N=100,000 for the simulation study, two situations are
presented by considering the proportion of cure in the population of 0.3 and 0.5. In
these cases, the censored proportions are observed in different levels.

In pursuance to find the maximization of the log-likelihood function, described in the

equation (11), the package called maxLik available in R developed by Henningsen and
Toomet (2011) was used. The numerical results are well-behaved since was not found
numerical problems using the SANN method (Simulated-annealing), such as failure
evidence of convergence or end on multiple maxima. The programs can be obtained,
upon request.

Table 1: MREs, MSEs, 𝐶𝐶95% estimates for 100,000 considering n = (25, 50, 100, 200,
300) and 45.7% of censorship.
𝜽𝜽 𝛼𝛼 = 0.5 𝜆𝜆 = 2.0 𝑝𝑝 = 0.3 0.457
𝑛𝑛 MRE MSE C95% MRE MSE C95% MRE MSE C95% 𝑀𝑀𝑝𝑝
25 1.265 0.060 0.948 1.100 3.344 0.810 1.100 0.021 0.925 0.461
50 1.114 0.020 0.952 1.117 2.002 0.860 1.024 0.013 0.938 0.458
100 1.048 0.008 0.952 1.098 1.087 0.893 0.992 0.008 0.948 0.457
200 1.022 0.004 0.953 1.059 0.488 0.919 0.991 0.004 0.952 0.457
300 1.014 0.003 0.951 1.039 0.293 0.928 0.993 0.003 0.952 0.457

Table 2: MREs, MSEs, 𝐶𝐶95% estimates for 100,000 considering n = (25, 50, 100, 200,
300) and 61% of censorship.
𝛼𝛼 = 0.5 𝜆𝜆 = 2.0 𝑝𝑝 = 0.5 0.612
n MRE MSE C95% MRE MSE C95% MRE MSE C95% 𝑀𝑀𝑝𝑝
25 1.384 0.144 0.939 1.263 8.714 0.788 1.027 0.022 0.914 0.612
50 1.158 0.033 0.940 1.238 5.967 0.838 1.001 0.014 0.928 0.612
100 1.070 0.013 0.946 1.156 2.683 0.877 0.994 0.007 0.940 0.612
200 1.031 0.006 0.951 1.085 0.841 0.906 0.994 0.004 0.948 0.612
300 1.020 0.004 0.951 1.056 0.459 0.921 0.996 0.002 0.951 0.612

7
Table 3: MREs, MSEs, 𝐶𝐶95% estimates for 100,000 considering n = (25, 50, 100, 200,
300) and 35% of censorship.
𝛼𝛼 = 2.0 𝜆𝜆 = 4.0 𝑝𝑝 = 0.3 0.35
n MRE MSE C95% MRE MSE C95% MRE MSE C95% 𝑀𝑀𝑝𝑝
25 1.103 0.316 0.951 1.022 0.347 0.921 0.997 0.010 0.927 0.349
50 1.047 0.116 0.951 1.011 0.155 0.938 0.998 0.005 0.937 0.348
100 1.023 0.050 0.950 1.005 0.074 0.945 1.000 0.002 0.945 0.349
200 1.011 0.023 0.950 1.003 0.036 0.947 1.000 0.001 0.947 0.349
300 1.008 0.015 0.951 1.002 0.024 0.947 0.999 0.001 0.947 0.348

Table 4: MREs, MSEs, 𝐶𝐶95% estimates for 100,000 considering n = (25, 50, 100, 200,
300) and 53.5% of censorship.
𝛼𝛼 = 2.0 𝜆𝜆 = 4.0 𝑝𝑝 = 0.3 0.535
n MRE MSE C95% MRE MSE C95% MRE MSE C95% 𝑀𝑀𝑝𝑝
25 1.158 0.619 0.953 1.034 0.546 0.910 0.998 0.011 0.933 0.535
50 1.068 0.182 0.953 1.016 0.230 0.933 0.999 0.006 0.942 0.535
100 1.033 0.075 0.950 1.007 0.105 0.942 1.000 0.003 0.947 0.535
200 1.016 0.034 0.950 1.004 0.051 0.946 0.999 0.001 0.947 0.534
300 1.011 0.022 0.949 1.002 0.034 0.948 1.000 0.001 0.948 0.535

The estimates obtained from Tables 1-4 for 𝜶𝜶, 𝝀𝝀 and 𝒑𝒑 are asymptotically unbiased,
implying that MREs tend to one when n increases and the MSEs decrease to zero for n
large. Analyzing the MLEs performance, with a coverage probabilities tending to 0.95,
good coverage properties may be deliberated for the parameter estimators. In practical
applications, those estimation procedures will be relevant as shown in the next section.

5 Application

In this section, we considered the data set presented by Kersey et al. (1987). The results
were collected in a group of 46 patients, per years, upon the recurrence of leukemia
whom received autologous marrow. Table 5 shows the full data set (+ indicates
censored observations).

8
Table 5: Leukemia free-survival times (in years) for the 46 autologous transplant
patients (where + indicates censored observations).
0.0301 0.0384 0.0630 0.0849 0.0877 0.0959 0.1397 0.1616
0.1699 0.2137 0.2137 0.2164 0.2384 0.2712 0.2740 0.3863
0.4384 0.4548 0.5918 0.6000 0.6438 0.6849 0.7397 0.8575
0.9096 0.9644 1.0082 1.2822 1.3452 1.4000 1.5260 1.7205+
1.9890+ 2.2438 2.5068+ 2.6466+ 3.0384 3.1726+ 3.4411 4.4219+
4.4356+ 4.5863+ 4.6904+ 4.7808+ 4.9863+ 5.0000+

The proposed model is compared with some usual long-term survival models, such as
the LT Weibull and LT weighted Lindley (Louzada and Ramos, 2017). Different
discrimination criterion methods are considered: the negative of the maximum value of
� ; 𝒕𝒕�, the Akaike information criterion (𝐴𝐴𝐴𝐴𝐴𝐴 = −2𝑙𝑙�𝜽𝜽
the likelihood function 𝑙𝑙�𝜽𝜽 � ; 𝒕𝒕� +

2𝑘𝑘) and the corrected AIC (𝐴𝐴𝐴𝐴𝐴𝐴 + 2𝑘𝑘(𝑘𝑘 + 1)/ (𝑛𝑛 − 𝑘𝑘 − 1)), where k is the number of
parameters to be fitted. The best model is the one which provides the minimum criterion
method values. Figure 2 presents the empirical survival function adjusted by the
Kaplan-Meier estimator and different LT survival distributions.

Figure 2: Survival function adjusted by the empirical survival function (Kaplan-Meier

estimator), LT Fréchet, LT Weibull and LT WL distribution.

Table 6 presents the results of the different discrimination criteria for different
probability distributions. Comparing the results of the different discrimination methods,

9
we observed that the LT Fréchet distribution has better fit then the LT models under the
Weibull and weighted Lindley baseline distribution.

Method LT Fréchet LT Weibull LT WL

− log 𝐿𝐿 45.33 46.15 46.56
AIC 96.66 98.30 99.12
AICc 97.23 98.87 99.69

The MLEs were obtained through the same procedure as described in Section 3. The
standard error (SE) and the confidence intervals, considering a 95% confidence level for
𝛼𝛼, 𝜆𝜆 and 𝑝𝑝 are displays in Table 7.

Table 7: MLE, Standard Error (SE), and confidence interval under 95% confidence
level for 𝛼𝛼, 𝜆𝜆 and 𝑝𝑝.

𝜽𝜽 MLE SE 𝐶𝐶𝐼𝐼95% (𝜽𝜽)

𝛼𝛼 0.65682 0.01975 (0.38140; 0.93225)
𝜆𝜆 0.31358 0.01531 ( 0.07106; 0.55609)
𝑝𝑝 0.12476 0.01597 (0.00000; 0.37245)

Note that, in Kersey et al. (1987) they use the non-parametric KM estimate of
the cure fraction in which was 0.20where (0.08; 0.32)is the 95% confidence interval.
Therefore, results showed to be consistence with Kersey et al. results while our estimate
was contained in the non-parametric interval. By using our parametric model the
estimate obtained for 𝑝𝑝 was 0.125 showing an overestimation of the long term survival
patients. As it can be seen, through our proposed methodology the data related to the
leukemia free-survival times (in months) for the 50 autologous transplant patients can
be described by the LF distribution.

6 Discussion

In this paper, we have proposed a new long-term survival distribution called long term
Fréchet distribution and its mathematical properties were studied. It was presented
results towards the maximum likelihood parameters’ estimators and their asymptotic
properties. The estimators’ efficient were present in the simulation study as the MLEs
for the three unknown parameters obtained acceptable results even for small sample
sizes. As such of the real dataset problem, related to the leukemia, free-survival times

10
(in months) for the 50 autologous transplant patients. Many extensions from this present
work can be considered, for instance, the parameters estimation may also be studied
under an objective Bayesian analysis (Ramos et al., 2014, 2017) or using different
classical methods (Louzada et al., 2016; Bakouch et al. 2017). Other approach should be
to include covariates under the assumption of Cox model, i.e., proportional hazards. In
conclusion, this regression model can be extended for the Bayesian approach as well.

Acknowledgements

The authors are thankful to the Editorial Board and to the reviewers for their valuable
comments and suggestions which led to this improved version.

References

Bakouch, H. S., Dey, S., Ramos, P. L., & Louzada, F. (2017). Binomial-exponential 2
Distribution: Different Estimation Methods and Weather Applications. Trends in
Applied and Computational Mathematics, 18(2), 233.

Barreto-Souza, W., Cordeiro, G. M., &Simas, A. B. (2011). Some results for beta
Fréchet distribution. Communications in Statistics-Theory and Methods, 40(5), 798-
811.

Berkson, J., & Gage, R. P. (1952). Survival curve for cancer patients following
treatment.Journal of the American Statistical Association, 47(259), 501-515.

De Gusmão, Felipe RS, Edwin MM Ortega, and Gauss M. Cordeiro."The generalized

inverse Weibull distribution." Statistical Papers 52.3 (2011): 591-619.

Jenkinson, A. F. (1955). The frequency distribution of the annual maximum (or

minimum) values of meteorological elements. Quarterly Journal of the Royal
Meteorological Society, 81(348), 158-171.

11
Henningsen, A., &Toomet, O. (2011). maxLik: A package for maximum likelihood
estimation in R. Computational Statistics, 26(3), 443-458.

Kersey, J. H., Weisdorf, D., Nesbit, M. E., LeBien, T. W., Woods, W. G., McGlave, P.
B., Kim, T., Vallera, D. A., Goldman, A. I., Bostrom, B. and Ramsay, N. K. C. (1987).
Comparison of autologous and autologous bone marrow transplantation for treatment of
high-risk refractory acutelymphoblastic leukemia. New England Journal of Medicine
317, 461-467.

Krishna, E., Jose, K. K., Alice, T., & Ristić, M. M. (2013).The Marshall-Olkin Fréchet
distribution.Communications in Statistics-Theory and Methods, 42(22), 4091-4107.

Louzada, F., Ramos, P. L.A New Long-Term Survival Distribution. Biostat Biometrics
Open Acc J. 1(4): 555574.

Louzada, F, Ramos, P.L and Perdoná.G. C. (2016).Different estimation procedures for

the parameters of the extended exponential geometric distribution for medical data.
Computational and mathematical methods in medicine, v.2016, 1-12.

Maller, R. A., & Zhou, S. (1995). Testing for the presence of immune or cured
individuals in censored survival data. Biometrics, 1197-1205.

Migon, H. S., Gamerman, D., & Louzada, F. (2014). Statistical inference: an integrated
approach. CRC press.

Perdoná, G. C., & Louzada-Neto, F. (2011).A general hazard model for lifetime data in
the presence of cure rate. Journal of Applied Statistics, 38(7), 1395-1405.

Ramos, P. L., Moala, F. A., & Achcar, J. A. (2014). Objective priors for estimation of
extended exponential geometric distribution. Journal of Modern Applied Statistical
Methods, 13(1), 226-243.

Ramos, P. L., Achcar, J.A., Moala, F. A., Ramos, E. & Louzada F., (2017) Bayesian
analysis of the generalized gamma distribution using non-informative priors, Statistics
51(4), 824-843.

12
Roman, M., Louzada, F., Cancho, V. G., & Leite, J. G. (2012).A new long-term
survival distribution for cancer data. Journal of Data Science, 10(2), 241-258.

13