A weighted random survival forest

Lev V. Utkin1 , Andrei V. Konstantinov1 , Viacheslav S. Chukanov1 ,

Mikhail V. Kots1 , Mikhail A. Ryabinin1 , Anna A. Meldo1,2
arXiv:1901.00213v1 [stat.ML] 1 Jan 2019

1
Peter the Great St.Petersburg Polytechnic University (SPbPU)
2
St.Petersburg Clinical Research Center for Special Types of Medical Care (Oncology-oriented)

Abstract
A weighted random survival forest is presented in the paper. It can be regarded
as a modification of the random forest improving its performance. The main idea
underlying the proposed model is to replace the standard procedure of averaging used
for estimation of the random survival forest hazard function by weighted avaraging
where the weights are assigned to every tree and can be veiwed as training paremeters
which are computed in an optimal way by solving a standard quadratic optimization
problem maximizing Harrell’s C-index. Numerical examples with real data illustrate
the outperformance of the proposed model in comparison with the original random
survival forest.
Keywords: random forest, decision tree, quadratic programming, survival analysis,
Harrell’s C-index, cumulative hazard function.

1 Introduction
A lot of computer aided diagnosis (CAD) systems have been developed in order to provide
successful detection of a disease and to facilitate making decision to start treatment process
at early stage. Most CAD systems aim to detect a disease or its features. However, there
are a few systems which take into account survival aspects of a patient especially of a
cancer patient. Two reasons of such the situation can be pointed out. First, CAD systems
taking into account survival aspects require the corresponding datasets which are mainly
lack or of a small size nowadays. Second, CAD systems have to handle data with censored
observations. This peculiarity may significantly complicate training process, and it requires
special methods for dealing with censored data. A large amount of structured data, which
has been recorded about patients, their peculiarities, do not take into account survival
aspects. Therefore, it is topical to develop models which could efficiently process the
available survival datasets in order to be an element of CAD systems.
A basis for such the models may be survival analysis or time-to-event analysis which can
be regarded as a fundamental tool which is used in many applied areas. One of the most

1
important areas is the medical research where survival models are widely used to evaluate
the significance of prognostic variables in outcomes such as death or cancer recurrence and
subsequently inform patients of their treatment options [34]. The datasets used in the
survival analysis or just the survival data differ from many datasets by the fact that time
to event of interest for a part of observations or instances is unknown because the event
might not have happened during the period of study [49]. If the observed survival time is
less than or equal to the true survival time, then we have a special case of censoring data
called right-censoring data. Other special cases are left-censoring and interval censoring
observations [65]. However, right-censoring is the most common case in many applications
[24]. Without loss of generality, we describe the survival models mainly in the medical
application terms below, i.e., instances will be called patients.
The survival models can be divided into three parts: parametric, nonparametric and
semiparametric. It is assumed in parametric models that the type of the probability distri-
bution of survival times is known, for example, the exponential, Weibull, normal, gamma
distributions. As pointed out by Lee and Wang [41], nonparametric or distribution-free
models are less efficient than parametric methods when survival times follow a theoretical
distribution and more efficient when no suitable theoretical distributions are known. They
can be used to analyze survival data before attempting to fit a theoretical distribution. One
of the simplest survivor models is the Kaplan-Meier estimator which is a non-parametric
model used to compute the survival function of a homogeneous data set. In other words,
the model does not take into account the fact that the instances may differ by their features.
A few critical features of the Kaplan-Meier model are considered in [41, Chapter 4, Page
76]. Nevertheless, the Kaplan-Meier model provides a simple way to compute the survival
function of patients.
A popular regression model for the analysis of survival data is the well-known Cox
proportional hazards model, which is a semi-parametric model that calculates the effects
of observed covariates on the risk of an event occurring, for example, the death or failure
[12]. The Cox model is the most commonly used regression analysis approach for survival
data among semi-parametric survival models. The model does not require knowledge of the
underlying distribution. It differs significantly from other methods since it is built on the
proportional hazards assumption and employs partial likelihood for parameter estimation.
The proportional hazards assumption in the Cox model means that different patients have
hazard functions that are proportional, i.e., the ratio of the hazard functions for two patients
with different prognostic factors or covariates is a constant and does not vary with time.
In other words, the ratio of the risk of dying of two patients is the same no matter how
long they survive [41, Chapter 12]. The model assumes that a patient’s log-risk of failure
is a linear combination of the patient’s covariates. This assumption is referred to as the
linear proportional hazards condition. It is interesting to note that the Cox model is
semi-parametric in the sense that it can be factored into a parametric part consisting of
a regression parameter vector associated with the covariates and a non-parametric part
that can be left completely unspecified [15]. Another interpretation of the semi-parametric
property of the Cox model is that we do not require to know the underlying distribution of

2
time to event of interest, but the attributes are assumed to have an exponential influence
on the outcome [65].
The Cox model is a very powerful method for dealing with survival data. As a result, a
lot of approaches dealing with the Cox model and its modifications have been proposed last
decades. A clear taxonomy of survival analysis methods and their comprehensive review
are presented by Wang et al. [65].
It should be noted that the Cox model may provide unsatisfactory results under condi-
tions of a high dimensionality of survivor data and a small number of observations. These
conditions take place in many application problems, for example, when we deal with gene
expression data. However, due to the high dimensionality of gene expression data when
the number of genes expressed exceeds the number of patients, it is not possible to take an
estimation approach based on the standard Cox model. To overcome this problem, Tib-
shirani [63] proposed one of the interesting modifications of the Cox model based on the
Lasso method. Kim et al. [38] considered the Cox regression with the group Lasso penalty
which improves the combination of different covariates, for example, clinical and genomic
covariates. The adaptive Lasso for the Cox model is proposed by [74]. Some modifications
of the Cox model with using Lasso can also be found in [16, 33, 39, 62, 67].
One of the main problems of the Cox model is linear relationship assumption between
covariates and the time of event occurrence. Various modifications have been proposed to
generalize the Cox model taking into account the corresponding non-linear relationship be-
tween covariates and the time of event. The first class of models uses a neural network for
modelling the non-linear function. Faraggi and Simon in their pioneering work [17] present-
ed an approach to modelling survival data using the input-output relationship associated
with a simple feed-forward neural network as the basis for a non-linear proportional haz-
ards model. The proposed model was a basis for developing more complex generalization
using the deep neural networks [34, 44, 49, 53, 71, 75]. The convolutional neural networks
(CNN) also have been applied to the survival analysis. In particular, Haarburger et al. [21]
used CNN for analysis of lung cancer patients and illustrated that the CNN improves the
predictive accuracy of Cox models that otherwise only rely on radiomics features. Some
aspects of application of the survival analysis to medical diagnostic problems have been
discussed by Afshar at al. [2]. Several models based on neural networks are considered
in the review by Wang et al. [65]. A review of deep learning methods for dealing with
survival data is presented by Nezhad et al. [49]. The proposed generalizations have many
advantages, but there is an important disadvantage. The use of neural networks requires a
lot of survival data. This condition is violated in many applications. Therefore, Van Belle
et al. [5, 6] proposed to use SVM in order to enhance the model by the small amount of
training data. The SVM approach to survival analysis has been studied by several authors
[36, 52, 59, 4, 66].
Another approach for dealing with the limited survival data is to use survival trees and
the random survival forests (RSFs). As pointed out by Wang et al. [65], the splitting criteria
as one of the main concepts of decision trees differ for survival trees. The splitting criteria
can be divided into two classes: minimizing within-node homogeneity and maximizing

3
between-node heterogeneity. The first class of approaches minimizes the loss function
using the within-node homogeneity criterion. Criteria from the first class measure the
within-node homogeneity with a statistic that measures how similar the subjects in each
node are and choose splits that minimize the within-node error. In particular, Gordon
and Olsen [20] proposed an extension of CART to survival data by applying a distance
measure, for example, the Wasserstein metric, between Kaplan-Meier curves and certain
point masses. Davis and Anderson [14] proposed another splitting criterion based on the
likelihood method under assumption that the survivor function in a node is exponential with
a constant hazard. An example of a splitting criterion from the second class is a criterion
using the log-rank test statistics presented by Ciampi [11]. Due to many advantages of
decision trees as a tool for classification and regression, several tree-based modifications
solving the survival analysis problem have been proposed last decades [27, 28, 40, 43, 58,
60, 72, 73]. Survival random forests have been applied to many real application problems,
for example, [3, 19, 46]. A detailed review of survival trees as well as RSFs is represented by
Bou-Hamad et al. [9]. A new algorithm for rule induction from survival data was proposed
by Wrobel et al. [70]. It works according to the separate-and-conquer heuristics with a use
of log-rank test for establishing rule body.
Random forests were introduced by Breiman [10] in order to overcome some shortcom-
ings of the decision trees, in particular, their instability to small perturbations in a learning
sample. The random forest uses a large number of randomly built patient decision trees in
order to combine their predictions. It also reduces the possible correlation between decision
trees by selecting different subsamples of the feature space.
It turns out that the random forests became a very powerful, efficient and popular tool
for the survival analysis. The random forest can be regarded as a nonparametric machine
learning strategy. The popularity of RSFs stems from many useful factors. First of all,
Ishwaran and Kogalur [30] pointed out that the random forests require only three tuning
parameters to be set (the number of randomly selected predictors, the number of trees
grown in the forest, and the splitting rule). Moreover, the random forest is highly data
adaptive and virtually model assumption free. Wang and Zhou [64] mention also that
random forests have proved to be successful in various scenarios including classification,
regression and survival analysis [7]. They can deal with both low and high dimensional data
while other popular ensembles often fail when confronted with high dimensional datasets.
As a result, a lot of models based on random forest have been developed for dealing with
survival data [8, 26, 29, 35, 47, 48, 50, 56, 61, 68, 69]. Most models are very similar and
differ in splitting criteria and the ensemble estimation. Splitting criteria totally defines the
survival trees in the random forest and has been briefly considered above. Most survival
random forests use averaging of the tree cumulative hazard estimates and its modifications.
It should be noted that other ensemble models and algorithms for dealing with survival
data have been developed, for example, Hothorn et al. [25] proposed a unified and flexible
framework for ensemble survival learning and introduced the corresponding random forest
and generic gradient boosting algorithms.
Since the RSF is one of the most efficient models in survival analysis, then we pay

4
attention to this model and propose an approach for its improving. The first idea underlying
the improvement is to modify the procedure of averaging used for estimation of the forest
survival function on the basis of survival functions of trees. We propose to replace the
standard averaging with the weighted sum of the tree survival functions. The corresponding
RSF with weights will be called weighted RSF (WRSF). By assigning the weights to every
tree survival function, we, in fact, assign these weights to every tree in the random forest
because the weights do not depend on the training examples. The second idea is that
weights in the sum are regarded as training parameters which can be computed in an
optimal way by solving an optimization problem. The third idea is to apply the concordance
error rate called C-index [22] for constructing the optimization problem. The C-index
estimates how good the model is at ranking survival times. It is one of the popular measures
for comparison survival models. It turns out that maximization of the C-index may be
a basis for training the tree weights. It should be noted that the use of the C-index
in its original form makes the optimization problem computationally hard to be solved.
Therefore, the fourth idea is to replace the C-index with its approximate representation
which is based on applying the well-known hinge loss function. As a result, we get the
standard quadratic optimization problem for computing optimal weights, which can be
solved by many available methods.
The weighting scheme in random forests is not new. Some random forest algorithms
assign weights to classes [13]. There are algorithms with weights of decision trees [37, 42, 54].
However, to the best of our knowledge, the weighting schemes have not been used in RSFs.
Moreover, in contrast to the available weighting algorithms in original random forests, the
proposed approach considers weights in the RSF as training parameters.

2 Some elements of survival analysis and a formal

problem statement
In survival analysis, a patient i is represented by a triplet (xi , δi , Ti ), where xi = (xi1 , ..., xim )
is the vector of the patient parameters (characteristics) or the vector of features; Ti indicates
time to event of the patient, it is assumed to be non-negative and continuous. If the event
of interest is observed, Ti corresponds to the time between baseline time and the time
of event happening, in this case δi = 1, and we have an uncensored observation. If the
instance event is not observed and its time to event is greater than the observation time, Ti
corresponds to the time between baseline time and end of the observation, and the event
indicator is δi = 0, and we have a censored observation. Suppose a training set D consists
of n triplets (xi , δi , Ti ), i = 1, ..., n. The goal of survival analysis is to estimate the time
to the event of interest T for a new patient with feature vector denoted by x by using the
training set D.
The survival and hazard functions are key concepts in survival analysis for describing
the distribution of event times. The survival function denoted by S(t) as a function of
time t is the probability of surviving up to that time, i.e., S(t) = Pr{T > t}. The hazard

5
function h(t) is the rate of event at time t given that no event occurred before time t, i.e.,

Pr{t ≤ T ≤ t + ∆t|T ≥ t} f (t)

h(t) = lim = ,
∆t→0 ∆t S(t)

where f (t) is the density function of the event of interest.

By using the fact that the density function can be expressed through the survival
function as
dS(t)
f (t) = − ,
dt
we can write the following expression for the hazard rate:
d
h(t) = − ln S(t).
dt
The survival function is determined through the hazard function as
 Z t 
S(t) = exp − h(z)dz = exp (−H(t)) ,
0

where H(t) is the cumulative hazard function.

We did not write the dependence of the above functions on a feature vector x for short.

2.1 The Cox model

According to the Cox proportional hazards model, [24], the hazard function at time t given
predictor values x is defined as

h(t|x) = h0 (t)Ψ(x, b) = h0 (t) exp (ψ(x, b)) .

Here h0 (t) is an arbitrary baseline hazard function; Ψ(x) is the covariate effect or the
risk function; b = (b1 , ..., bm ) is an unknown vector of regression coefficients or parameters.
It can be seen from the above expression for the hazard function that the reparametrization
Ψ(x, b) = exp (ψ(x, b)) is used in the Cox model. The function ψ(x, b) in the model is
linear, i.e., Xm
ψ(x, b) = xbT = bk x k .
k=1

In the framework of the Cox model, the survival function S(t) is computed as

S(t) = exp(−H0 (t) exp (ψ(x, b))) = (S0 (t))exp(ψ(x,b)) .

Here H0 (t) is the cumulative baseline hazard function; S0 (t) is the baseline survival
function.

6
 The partial likelihood in this case is defined as follows:
n
" # δj
Y exp(ψ(xj , b))
L(b) = P .
j=1 i∈Rj exp(ψ(xi , b))

Here Rj is the set of patients who are at risk at time tj . The term “at risk at time t”
means patients who die at time t or later.
It should be note that the idea underlying the use of neural networks in survival analysis
is to replace the linear function ψ(x) with a non-linear function which is realized by means
of a neural network [17].
In order to provide personalized treatment recommendations in accordance with the
recommender function, we compute the functions ψi (x) and ψj (x) corresponding to dif-
ferent treatment groups. If the obtained function recij (x) is positive, then treatment j is
preferable in comparison with treatment i. In the case of a negative recommender function,
treatment i is more effective and leads to a lower risk than treatment j (see, for example,
[34]).
To compare the survival models, the C-index proposed by Harrell et al. [22] is used.
The C-index estimates how good the model is at ranking survival times. It estimates the
probability that, in a randomly selected pair of patients, the patient that fails first had
a worst predicted outcome. In fact, this is the probability that the event times of a pair
of patients are correctly ranking. C-index does not depend on choosing a fixed time for
evaluation of the model and takes into account censoring of patients [45].
Let us consider the training set D consisting of n triplets (xi , δi , Ti ). We consider possible
or admissible pairs {(xi , δi , Ti ), (xj , δj , Tj )} for i ≤ j. Then the C-index is calculated as
the ration of the number of pairs correctly ordered by the model to the total number of
admissible pairs. A pair is not admissible if the events are both right-censored or if the
earliest time in the pair is censored. If the C-index is equal to 1, then the corresponding
survival model is supposed to be perfect. If the C-index is 0.5, then the model is no better
than random guessing.
Let t∗1 , ..., t∗q denote predefined time points, for example, t1 , ..., tN , where N is distinct
event times. If the output of a survival algorithm is the predicted survival function S(t),
then the C-index is formally calculated as [65]:
1 X X  ∗
1 S(ti |xi ) > S(t∗j |xj ) .

C= (1)
M i:δ =1 j:t <t
i i j

Here M is the number of all comparable or admissible pairs; 1[a] is the indicator function
taking the value 1 if a is true, and 0 otherwise; S is the estimated survival function.
It should be noted that there are different definitions of the C-index, which depend on
the output of a survival algorithm. However, we will use the definition (1) which plays an
important role in the proposed improvement of the RSF.

7
2.2 Random survival forests
It has been mentioned that the RSF is one of the best models for survival analysis due
to its properties. This is the main reason for its modifying below in order to improve the
survival analysis results and to increase the prediction accuracy.
A general algorithm of constructing RSFs can be represented as follows [31]:
1. Draw Q bootstrap samples from the original data. Note that each bootstrap sample
excludes on average 37% of the data, called out-of-bag data (OOB data).
2. Grow √ a survival tree for each bootstrap sample. At each node of the tree, ran-
domly select m candidate variables. The node is split using the candidate variable that
maximizes survival difference between daughter nodes.
3. Grow the tree to full size under the constraint that a terminal node should have no
less than d > 0 unique deaths.
4. Calculate a cumulative hazard function for each tree or a survival function. Average
to obtain the ensemble cumulative hazard function or the ensemble survival function.
5. Using out-of-bag data, calculate prediction error for the ensemble cumulative hazard
function or the ensemble survival function.
The parameters of the algorithm proposed by Ishwaran et al. [31] and some its steps
may vary, but, generally, it can be viewed as a basis for solving the survival analysis problem
by means of many its implementations and modifications.
The most important question of the RSFs, which defines their different implementations
is the splitting rule. As shown by Ishwaran et al. [31], a good split maximizes survival
difference across the two sets of data. We shortly review the main splitting rules used in
RSF [31, 65].
Let t1 < t2 < ... < tN be the distinct times to event of interest, for example, times to
deaths, in the parent node g, and let Zij and Yij equal the number of deaths and patients
at risk at time ti in the daughter nodes j = 1, 2, i.e.,

Yi1 = #{Tl ≥ ti , xl ≤ c}, Yi2 = #{Tl ≥ ti , xl > c}.

Here xl is the value of a feature x for the l-th patient, l = 1, ..., n. Let Yi = Yi1 + Yi2
and Zi = Zi1 + Zi2 . Let n1 and n2 be total numbers of observations in daughter nodes such
that n = n1 + n2 , i.e.,

n1 = #{l : xl ≤ c}, n2 = #{l : xl > c}.

The log-rank test for a split at the value c for predictor x is defined as
PN
i=1 (Zi1 − Yi1 Zi /Yi )
L(x, c) = r  .
PN Yi1  Yi1

Yi −Zi
i=1 Yi 1 − Yi Yi −1
Zi

The value |L(x, c)| is the measure of node separation, which should be minimized for
better splitting.

8
 An idea underlying another splitting rule called as conservation of events splitting is
to suppose that the sum of estimated cumulative hazard functions over the observed time
points must equal the total number of deaths. By using the notations introduced for the
log-rank test, the measure of conservation of events for the split on x at the value c can be
defined as
−1
2 N
( k
)
1 X X X Zlj
Cons(x, c) = Y1j Nkj Yk+1,j .
Y11 + Y12 j=1 k=1 l=1
Ylj
It should be noted that the splitting rule should maximize survival differences due to the
split. Therefore, the transformed value 1/(1 + Cons(x, c)) as a measure of node separation
is used. PN
We also consider the approximate log-rank splitting. Let Z = i=1 Zi and Z1 =
PN
i=1 Zi1 . The log-rank test L(x, c) is

Z 1/2 (Z1 − nl=1 1 {xl ≤ c} H(Tl ))

P
L(x, c) = p Pn .
( l=1 1 {xl ≤ c} H(Tl )) (Z − nl=1 1 {xl ≤ c} H(Tl ))
P

The next important question is how to compute the ensemble hazard function or the
ensemble survival function. First, we consider how to compute the cumulative hazard
estimate for the k-th terminal node of a tree. Let {tj,k } be the N (k) distinct death times in
terminal node k of the q-th tree such that t1,k < t2,k < ... < tN (k),k and Zj,k and Yj,k equal
the number of deaths and patients at risk at time tj,k . The cumulative hazard estimate for
node k is defined as (the Nelson–Aalen estimator):
X
Hk (t) = Zj,k /Yj,k .
tj,k ≤t

If the i-th patient with features xi falls into node k, then we can say that H(t|xi ) =
Hk (t). The ensemble cumulative hazard estimate for the i-th patient is obtained by aver-
aging cumulative hazard estimates of all Q trees, i.e.,
Q
1 X
Hf (t|xi ) = Hq (t|xi ). (2)
Q q=1

The survival function can be obtained from Hq (t|xi ) as follows:

Sq (t|xi ) = exp (−Hq (t|xi )) .
Another ensemble estimate is considered by Ishwaran et al. [31], where OOB data are
used. Let Oq be a set of OBB example indexes for the tree q. The OOB prediction for each
training example xi uses only the trees that did not have xi in their bootstrap sample. If
we define the indicator function as 1(i ∈ Oq ), then the OOB ensemble cumulative hazard
estimator for the i-th training example is defined as
PQ
q=1 1(i ∈ Oq ) · Hq (t|xi )
Hf (t|xi ) = PQ . (3)
q=1 1(i ∈ Oq )

9
3 Weights of survival decision trees
One can see from (2) that the ensemble cumulative hazard estimate Hf (t|xi ) is obtained
under condition that all trees have the same weights 1/Q. A straightforward way to improve
the random forest is to assign weights w = (w1 , ..., wQ ) to decision trees. At that, it is
assumed that the sum of weight is 1, i.e., every vector w belongs to the unit simplex of the
dimension Q. As a result, we replace the averaging of the cumulative hazard estimates (2)
by weighted averaging for computing the the cumulative hazard function as follows:
Q
X
Hf (t, w|xi ) = wq Hq (t|xi ). (4)
q=1

One of the ways for assigning the weights is to suppose that they are training parameters
which can be optimized in accordance with a goal. Therefore, we have to define the goal
or an objective function for getting optimal weights.
One of the most important measure for comparison different models is the C-index
defined in (1). If we assume that the predicted survival function of the random forest
depends on the weights, we can maximize the C-index with respect to the weights. Let us
write the C-index as a function of the weights
1 X X 
1 Sf (t∗i , w|xi ) − Sf (t∗j , w|xj ) > 0 .

C(w) = (5)
M i:δ =1 j:t <t
i i j

Here Sf (t∗i , w|xi ) is the ensemble predicted survival function depending on weights w
of trees. By maximizing the C(w) over the non-negative weights wq , q = 1, ..., Q, under
constraint Q
P
q=1 wq = 1, we can get optimal weights.
It is difficult to solve the optimization problem with the indicator function in the objec-
tive function (5) because we have a hard combinatorial problem. Moreover, the dependence
of the ensemble survivor function on the weights is non-linear because
Q
!
X
Sf (t, w|xi ) = exp (−Hf (t, w|xi )) = exp − wq Hq (t|xi ) .
q=1

Fortunately, we can overcome this difficulty as follows. Note that there holds

1 Sf (t∗i , w|xi ) > Sf (t∗j , w|xj )

 

= 1 ln Sf (t∗i , w|xi ) > ln Sf (t∗j , w|xj )

 

= 1 Hf (t∗j , w|xi ) > Hf (t∗i , w|xj ) .

 

Hence, we get
1 X X 
1 Hf (t∗j , w|xi ) − Hf (t∗i , w|xj ) > 0 .

C(w) =
M i:δ =1 j:t <t
i i j

10
 Let us denote the set of all possible pairs (i, j) in (5), satisfying the condition δi = 1
for i and the condition ti < tj for j, as J. Taking into account (4), we get the following
optimization problem:
" Q #
1 X X
wq Hq (t∗j |xj ) − Hq (t∗i |xi ) > 0 ,


C(w) = max 1 (6)
w M
(i,j)∈J q=1

subject to
Q
X
wq = 1, wq ≥ 0, q = 1, ..., Q. (7)
q=1

The constraints for weights produce the unit simplex denoted as ∆Q whose dimension-
ality is Q. By maximizing C(w) over w ∈ ∆Q , we can get optimal weights.
One of the obvious ways for simplifying the optimization problem is to replace the
indicator function with the sigmoid σ, i.e., the optimization problem becomes to be
" Q #
1 X X
wq Hq (t∗j |xj ) − Hq (t∗i |xi ) .


C(w) = max σ (8)
w∈∆Q M
(i,j)∈J q=1

It can be seen from the objective function that the problem can be solved by applying
the gradient descent method. However, the main difficulty here is to take into account the
linear constraints for weights (7) which can be represented as the unit simplex of weights
denoted as ∆Q whose dimensionality is Q.
Another way for simplifying the optimization problem is to replace the indicator function
with the hinge loss function similarly to the replacement proposed by Van Belle et al. [5].
The hinge loss function is of the form:

l(x) = max (0, x) .

By adding the regularization term, we can write the optimization problem as

 
Q
!
X X 
∗ ∗


min max 0, wq Hq (ti |xi ) − Hq (tj |xj ) + λR(w) . (9)
w∈∆Q  
(i,j)∈J q=1

Here R(w) is a regularization term, λ is a hyper-parameter which controls the strength

of the regularization. We define the regularization term as

R(w) = kwk2 .

Let us introduce the variables

Q
!
X
wq Hq (t∗i |xi ) − Hq (t∗j |xj )


ξij = max 0, . (10)
q=1

11
Then the optimization problem is of the form:
 
X 
2
min ξij + λ kwk , (11)
w  
(i,j)∈J

subject to w ∈ ∆Q and
Q
X
wq Hq (t∗i |xi ) − Hq (t∗j |xj ) , ξij ≥ 0, {i, j} ∈ J.


ξij ≥ (12)
q=1

We get a standard quadratic optimization problem with linear constraints and with the
vector w of Q variables. It can be solved by many known methods.
It is interesting to note that the above optimization problem is very similar to the primal
form of the well-known SVM [57].
A general algorithm for training the WRSF taking into account weighted ensemble
estimation can be regarded as an extension of the algorithm given in previous section for
the original RSF. Given the training set D = {(xi , δi , Ti ), i = 1, ..., n}, xi ∈ Rm , δi ∈
{0, 1}, Ti ∈ R, we use the cumulative hazard functions Hq (t∗i |xi ) of all trees (q = 1, ..., Q)
corresponding to the i-th example, i = 1, ..., n, and solve the optimization problem (11)-
(12). Taking the optimal weights w as the solution of (11)-(12), we use (4) in order to get
the ensemble survival function.
It is interesting to note that the above optimization problem is very similar to the
primal form of the SVM modification for survival analysis [5, Problem (11)]. Indeed,
the objective
P functions are identical. Constraints in the survival SVM are of the form:
ξij ≥ 1+ m k=1 wq (xik − xjk ). One can see that the idea of the SVM modification for survival
analysis is to find a line which separates ranking points xi − xj . By using the problem (11)-
(12), we try to find a line which separates the ranking points H(t∗i |xi ) − H(t∗j |xj ), where
H(t|x) is the vector of the cumulative hazard function estimates for all trees at time t
by testing x. If the SVM modification for survival analysis deals with pairs of feature
vectors, then the proposed WRSF analyses pairs of the decision tree outputs. From this
point of view, the proposed procedure for training the weights of trees can be regarded as
a second-order SVM or meta-learner for the RSF.
It should be noted that the number of weights is equal to the number of trees in
the forest. On the one hand, we would like to improve the classification algorithm by
introducing the weights. On the other hand, we get a lot of training parameters which may
lead to overfitting by a small amount of training data. In order to overcome this difficulty,
we propose to reduce the number of weights by grouping trees into identical subsets and by
assigned weights to the subsets. Suppose that we divide all trees into G subsets such that
every subset consists of g trees, g · G = Q. Then we have G weights and the optimization
problem (11)-(12) can be rewritten as
 
X 
2
min ξij + λ kwk , (13)
w  
(i,j)∈J

12
subject to w ∈ ∆G and
G
X  
ξij ≥ wk H e k (t∗j |xj ) , ξij ≥ 0, {i, j} ∈ J.
e k (t∗i |xi ) − H (14)
k=1

Here He k (t∗ |x) is the mean cumulative hazard function of the k-th subset of trees. The
parameters G and g can be regarded as tuning parameters in place of the parameter Q.
Another difficulty of solving the optimization problem (11)-(12) is a large number of
constraints for ξij because all admissible pairs of training data with indices from the set J
produce them. It is interesting to point out that the same difficulty has been considered
in the SVM modification for survival analysis [6] where a scalable nearest neighbor algo-
rithm was proposed to reduce computational load without considerable loss of performance.
According to this algorithm, the number of constraints can be reduced by selecting a set
Ji of k samples with a survival time nearest to the survival time of sample i. However,
we use another approach. In order to simplify the optimization problem, we propose to
reduce the number of constraints by random selection of K constraints from the whole set
of constraints which is defined by all pairs of indices in the set J. Of course, we may get
a non-optimal solution in this case. However, our numerical experiments have shown that
this simplification of the optimization problem provides better results than the original
RSF.

4 Numerical experiments
Since the WRSF can be viewed as an improvement of the original RSF, then our interest
in this study is to compare the weighted RSF and the original RSF.
In order to carry out the comparisons, the proposed weighted RSF is tested on seven
real benchmark datasets. A short introduction of the benchmark datasets are given below.
The Primary Biliary Cirrhosis (PBC) Dataset contains observations of 418 pa-
tients with primary biliary cirrhosis of the liver from the Mayo Clinic trial [18], 257 of
whom have censored data. Every example is characterized by 17 features including age,
sex, ascites, hepatom, spiders, edema, bili and chol, etc. The dataset can be obtained via
the “randomForestSRC” R package.
The German Breast Cancer Study Group 2 (GBSG2) Dataset contains obser-
vations of 686 women [55]. Every example is characterized by 10 features, including age
of the patients in years, menopausal status, tumor size, tumor grade, number of positive
nodes, hormonal therapy, progesterone receptor, estrogen receptor, recurrence free survival
time, censoring indicator (0 - censored, 1 - event). The dataset can be obtained via the
“TH.data” R package.
The Chronic Myelogenous Leukemia Survival (CML) Dataset is simulated ac-
cording to structure of the data by the German CML Study Group used in [23]. The
dataset consists of 507 observations with 7 feature: a factor with 54 levels indicating the

13
study center; a factor with levels trt1, trt2, trt3 indicating the treatment group; sex (0 =
female, 1 = male); age in years; risk group (0 = low, 1 = medium, 2 = high); censoring
status (FALSE = censored, TRUE = dead); time survival or censoring time in days. The
dataset can be obtained via the “multcomp” R package (cml).
The Bladder Cancer Dataset (BLCD) [51] (Chapter 21) contains da-
ta on 86 patients after surgery assigned to placebo or chemotherapy (thiope-
ta). Endpoint is time to recurrence in months. Data on the number of
tumors removed at surgery was also collected. The dataset is available at
http://www.stat.rice.edu/˜sneeley/STAT553/Datasets/survivaldata.txt.
The Lupus Nephritis Dataset (LND) [1] contains data on 87 persons
with lupus nephritis. followed for 15+ years after an initial renal biop-
sy (the starting point of follow-up). This data set only contains time to
death/censoring, indicator, duration and log(1+duration), where duration is the
duration of untreated disease prior to biopsy. The dataset is available at
http://www.stat.rice.edu/˜sneeley/STAT553/Datasets/survivaldata.txt.
The Heart Transplant Dataset (HTD) contains data on 69 patients receiving heart
transplants [32]. This dataset is available at http://lib.stat.cmu.edu/datasets/stanford.
The Veterans’ Administration Lung Cancer Study (Veteran) Dataset [32] con-
tains data on 137 males with advanced inoperable lung cancer. The subjects were randomly
assigned to either a standard chemotherapy treatment or a test chemotherapy treatment.
Several additional variables were also measured on the subjects. The dataset can be ob-
tained via the “survival” R package.
The WRSF uses a software in Python to implement the procedures for computing
optimal weights of trees, the corresponding C-index of the whole random forest, and oth-
er procedures required for training and testing the WRSF. The software is available at
https://github.com/andruekonst/weighted-random-survival-forest.
To evaluate the C-index, we perform a cross-validation with 100 repetitions, where in
each run, we randomly select 75% of data for training and 25% for testing. Different values
for the regularization hyper-parameter λ have been tested, choosing those leading to the
best results.
Table 1 summarizes the numerical results for RSF and WRSF by different datasets
(column 1). At that, Table 1 shows the mean values of the C-index (columns 2 and 3), the
standard deviation (Std) (columns 4 and 5) and the median of the C-index (columns 6 and
7). It can be seen from Table 1 that the WRSF outperforms the RSF for all datasets. It is
also interesting to point out that the standard deviation is decreased when we use WRSF.
We have mentioned that in the previous section that the number of trained weights may
lead to reduction of the WRSF performance due to overfitting. Therefore, it is interesting
to study how the C-index depends on the weight number. We take 500 trees and divide
them into 50, 100, 250 subsets such that every subset contains 10, 5, 2 trees, respectively.
Every subset of trees can be viewed as a small RSF and it has its trained weight. The
corresponding boxplots of the model performances by 50, 100, 250 weights for all datasets
are shown in Figs. 1-7. It turns out that the number of weights improves the WRSF

14
 Table 1: Comparison of the RSF and WRSF for different datasets
Mean value Std Median
Dataset RSF WRSF RSF WRSF RSF WRSF
PBC 0.888 0.910 0.013 0.010 0.889 0.911
GBSG2 0.889 0.910 0.013 0.010 0.891 0.911
BLCD 0.880 0.934 0.060 0.042 0.891 0.942
CML 0.889 0.910 0.013 0.010 0.889 0.911
LND 0.882 0.941 0.051 0.043 0.877 0.944
HTD 0.859 0.931 0.056 0.044 0.873 0.943
Veteran 0.870 0.929 0.046 0.041 0.882 0.943

Figure 1: The boxplot for the PBC dataset

performance for all datasets. Moreover, it is clearly seen from the boxplots that the WRSF
outperforms the RSF especially by large number of weights.
An interesting question is which values of weights are assigned to trees. In order to
answer this question, we provide a typical histogram of the weight values derived for the
dataset CML (see Fig. 8). The weights are sorted in the descending order. The largest
weight is 0.086, the smallest weight is 4 × 10−4 .
Another interesting question is how the model performance depends on the number
of trees in the random forest. The dependence of the C-index on the number of trees is
illustrated in Fig. 9 where the solid and dotted lines correspond to the RSF and the WRSF,
respectively. It can be seen from Fig. 9 that the large number of trees may lead even to
the performance deterioration when we use the RSF. Whereas the large number of trees
improves the WRSF.
Our next experiment aims to check whether we can apply the constraint reduction pro-

15
Figure 2: The boxplot for the GCSG2 dataset

Figure 3: The boxplot for the BLCD dataset

16
Figure 4: The boxplot for the CML dataset

Figure 5: The boxplot for the LND dataset

17
 Figure 6: The boxplot for the HTD dataset

Figure 7: The boxplot for the Veteran dataset

18
 Figure 8: A histogram of the weight values

Figure 9: Dependence of the C-index on the number of trees

19
 Figure 10: Dependence of the C-index on the number of selected constraints

cedure in order to simplify calculations. We reduce the number of constraints by random

selection of K constraints for ξij from the whole set of constraints which is defined by all
pairs of indices in the set J. We use again the Veteran dataset for experiments. Fig. 10
shows the dependence of the C-index on the number of selected constraints for optimiza-
tion. We can see from Fig. 10 that the C-index increases with the number of constraints.
Moreover, it is important to note that the C-index for RSF is less than the corresponding
C-index for WRSF. This implies that the number of constraints may be reduced in many
cases.

5 Conclusion
A new survival model based on using the weighted modification of the RSF has been
presented in the paper. The main idea underlying this model is to improve the RSF by
assigning weights to survival decision trees or to their subsets. The weights are viewed
as training parameters. It turns out that this approach provides very improved results
especially for some datasets, for example, BLCD, LND, HTD, Veteran. Numerical experi-
ments have illustrated that the proposed model may provide significantly better results in
comparison with the original RSF.
The proposed model has several advantages. First, the weights are assigned in ac-
cordance with the tree capability to correctly determine the cumulative hazard function.
Second, the weights are training parameters which are computed by solving the standard
quadratic optimization problem. As a results, the proposed approach is very simple. But
the main advantage of the model is that it opens a door for developing a controllable RSF
which can solve various machine learning problems in the framework of survival analysis,

20
including, for example, transfer learning. This can be done by changing the loss function
which depends on the weights. The consideration of these problems is a direction for further
research.
We have studied only the case of linear dependence of the C-index on the weights.
However, it is interesting to consider non-linear cases. One of the ways for implementing
this case is to use a neural network which is trained to maximize the obtained C-index.
The application of the neural network as an additional element of WRSF is also a direction
for further research.
Another problem of the WRSF as well as the RSF is that the number of cases when xi
falls into the k-th terminal node of a tree may be very small. It makes confidence bounds
for the Nelson–Aalen estimator, which estimates the cumulative hazard function, to be
very large. The development of robust models taking into account this problem is another
direction for further research.

Acknowledgement
This work is supported by the Russian Science Foundation under grant 18-11-00078.

References
[1] M. Abrahamowicz, T. MacKenzie, and J.M. Esdaile. Time-dependent hazard ratio:
modelling and hypothesis testing with application in lupus nephritis. JASA, 91:1432–
1439, 1996.

[2] P. Afshar, A. Mohammadi, K.N. Plataniotis, A. Oikonomou, and H. Benali. From

hand-crafted to deep learning-based cancer radiomics: Challenges and opportunities.
arXiv:1808.07954v1, Aug 2018.

[3] H. Akai, K. Yasaka, A. Kunimatsu, M. Nojima, T. Kokudo, N. Kokudo, K. Hasegawa,

O. Abe, K. Ohtomo, and S. Kiryu. Predicting prognosis of resected hepatocellular
carcinoma by radiomics analysis with random survival forest. Diagnostic and Inter-
ventional Imaging, 99(10):643–651, 2018.

[4] V. Van Belle, K. Pelckmans, S. Van Huffel, and J.A. Suykens. Support vector meth-
ods for survival analysis: a comparison between ranking and regression approaches.
Artificial intelligence in medicine, 53(2):107–118, 2011.

[5] V. Van Belle, K. Pelckmans, J.A.K. Suykens, and S. Van Huffel. Support vector
machines for survival analysis. In Proceedings of the Third International Conference
on Computational Intelligence in Medicine and Healthcare (CIMED2007), pages 1–8,
2007.

21
 [6] V. Van Belle, K. Pelckmans, J.A.K. Suykens, and S. Van Huffel. Survival svm: a
practical scalable algorithm. In ESANN, pages 89–94, 2008.

[7] G. Biau and E. Scornet. A random forest guided tour. Test, 25(2):197–227, 2016.

[8] I. Bou-Hamad, D. Larocque, and H. Ben-Ameur. Discrete-time survival trees and

forests with time-varying covariates: application to bankruptcy data. Statistical Mod-
elling, 11(5):429–446, 2011.

[9] I. Bou-Hamad, D. Larocque, and H. Ben-Ameur. A review of survival trees. Statistics

Surveys, 5:44–71, 2011.

[10] L. Breiman. Random forests. Machine learning, 45(1):5–32, 2001.

[11] A. Ciampi. Generalized regression trees. Computational Statistics & Data Analysis,
12:57–78, 1991.

[12] D.R. Cox. Regression models and life-tables. Journal of the Royal Statistical Society,
Series B (Methodological), 34(2):187–220, 1972.

[13] M.E.H. Daho, N. Settouti, M.E.A. Lazouni, and M.E.A. Chikh. Weighted vote for
trees aggregation in random forest. In 2014 International Conference on Multimedia
Computing and Systems (ICMCS), pages 438–443. IEEE, April 2014.

[14] R.B. Davis and J.R. Anderson. Exponential survival trees. Statistics in Medicine,
8(8):947–961, 1989.

[15] K. Devarajn and N. Ebrahimi. A semi-parametric generalization of the cox propor-

tional hazards regression model: Inference and applications. Computational Statistics
& Data Analysis, 55(1):667–676, 2011.

[16] J. Fan, Y. Feng, and Y. Wu. Borrowing Strength: Theory Powering Applications –
A Festschrift for Lawrence D. Brown, volume 6, chapter High-dimensional variable
selection for Cox’s proportional hazards model, pages 70–86. Institute of Mathematical
Statistics, 2010.

[17] D. Faraggi and R. Simon. A neural network model for survival data. Statistics in
medicine, 14(1):73–82, 1995.

[18] T.R. Fleming and D.P. Harrington. Counting processes and survival aalysis. John
Wiley & Sons, Hoboken, NJ, USA, 1991.

[19] J. Gilhodes, C. Zemmour, S. Ajana, A. Martinez, J.-P. Delord, E. Leconte, J.-M.

Boher, and T. Filleron. Comparison of variable selection methods for high-dimensional
survival data with competing events. Computers in Biology and Medicine, 91:159–167,
2017.

22
[20] L. Gordon and R.A. Olshen. Tree-structured survival analysis. Cancer treatment re-
ports, 69(10):1065–1069, 1985.
[21] C. Haarburger, P. Weitz, O. Rippel, and D. Merhof. Image-based survival analysis for
lung cancer patients using cnns. arXiv:1808.09679v1, Aug 2018.
[22] F. Harrell, R. Califf, D. Pryor, K. Lee, and R. Rosati. Evaluating the yield of medical
tests. Journal of the American Medical Association, 247:2543–2546, 1982.
[23] R. Hehlmann, H. Heimpel, J. Hasford, H.J. Kolb, H. Pralle, D.K. Hossfeld, W. Queiss-
er, H. Loeffler, A. Hochhaus, and B. Heinze. Randomized comparison of interferon-
alpha with busulfan and hydroxyurea in chronic myelogenous leukemia. the german
cml study group. Blood, 84(12):4064–4077, 1994.
[24] D. Hosmer, S. Lemeshow, and S. May. Applied Survival Analysis: Regression Modeling
of Time to Event Data. John Wiley & Sons, New Jersey, 2008.
[25] T. Hothorn, P. Bühlmann, S. Dudoit, A. Molinaro, and M.J. van der Laan. Survival
ensembles. Biostatistics, 7(3):355–373, 2006.
[26] C. Hu and J.A. Steingrimsson. Personalized risk prediction in clinical oncology re-
search: Applications and practical issues using survival trees and random forests.
Journal of Biopharmaceutical Statistics, 28(2):333–349, 2018.
[27] X. Huang, S. Chen, and S. Soong. Piecewise exponential survival trees with time-
dependent covariates. Biometrics, 54:1420–1433, 1998.
[28] N.A. Ibrahim, A. Kudus, I. Daud, and M.R. Abu Bakar. Decision tree for competing
risks survival probability in breast cancer study. International Journal Of Biological
and Medical Research, 3(1):25–29, 2008.
[29] H. Ishwaran, E.H. Blackstone, C.E. Pothier, and M.S. Lauer. Relative risk forests
for exercise heart rate recovery as a predictor of mortality. Journal of the American
Statistical Association, 99:591–600, 2004.
[30] H. Ishwaran and U.B. Kogalur. Random survival forests for r. R News, 7(2):25–31,
2007.
[31] H. Ishwaran, U.B. Kogalur, E.H. Blackstone, and M.S. Lauer. Random survival forests.
Annals of Applied Statistics, 2:841–860, 2008.
[32] J. Kalbfleisch and R. Prentice. The Statistical Analysis of Failure Time Data. John
Wiley and Sons, New York, 1980.
[33] S. Kaneko, A. Hirakawa, and C. Hamada. Enhancing the lasso approach for devel-
oping a survival prediction model based on gene expression data. Computational and
Mathematical Methods in Medicine, 2015(Article ID 259474):1–7, 2015.

23
[34] J.L. Katzman, U. Shaham, A. Cloninger, J. Bates, T. Jiang, and Y. Kluger. Deepsurv:
Personalized treatment recommender system using a Cox proportional hazards deep
neural network. BMC medical research methodology, 18(24):1–12, 2018.

[35] M. Khalilia, S. Chakraborty, and M. Popescu. Predicting disease risks from highly
imbalanced data using random forest. BMC Medical Informatics and Decision Making,
11(51):1–13, 2011.

[36] F.M. Khan and V.B. Zubek. Support vector regression for censored data (svrc): a
novel tool for survival analysis. In Proceedings of the IEEE International Conference
on Data Mining (ICDM), pages 863–868. IEEE, 2008.

[37] H. Kim, H. Kim, H. Moon, and H. Ahn. A weight-adjusted voting algorithm for
ensemble of classifiers. Journal of the Korean Statistical Society, 40(4):437–449, 2011.

[38] J. Kim, I. Sohn, S.-H. Jung, S. Kim, and C. Park. Analysis of survival data with group
lasso. Communications in Statistics - Simulation and Computation, 41(9):1593–1605,
2012.

[39] O. Krasotkina and V. Mottl. A bayesian approach to sparse Cox regression in high-
dimentional survival analysis. In Proceedings of the 11th International Conference on
Machine Learning and Data Mining in Pattern Recognition, volume 9166, pages 425–
437, New York, 2015. Springer-Verlag.

[40] M. LeBlanc and J. Crowley. Relative risk trees for censored survival data. Biometrics,
48(2):411–425, 1992.

[41] E.T. Lee and J.W. Wang. Statistical Methods for Survival Data Analysis. John Wiley
& Sons, New Jersey, 2003.

[42] H. B. Li, W. Wang, H. W. Ding, and J. Dong. Trees weighting random forest method
for classifying high-dimensional noisy data. In 2010 IEEE 7th International Conference
on E-Business Engineering, pages 160–163. IEEE, Nov 2010.

[43] A. Linden and P.R. Yarnold. Modeling time–to–event (survival) data using classifica-
tion tree analysis. Journal of Evaluation in Clinical Practice, 23(6):1299–1308, 2017.

[44] M. Luck, T. Sylvain, H. Cardinal, A. Lodi, and Y. Bengio. Deep learning for patient-
specific kidney graft survival analysis. arXiv:1705.10245, May 2017.

[45] M. May, P. Royston, M. Egger, A.C. Justice, and J.A.C. Sterne. Development and
validation of a prognostic model for survival time data: application to prognosis of HIV
positive patients treated with antiretroviral therapy. Statistics in Medicine, 23:2375–
2398, 2004.

24
[46] F. Miao, Y.-P. Cai, Y.-X. Zhang, Y. Li, and Y.-T. Zhang. Risk prediction of one-year
mortality in patients with cardiac arrhythmias using random survival fores. Computa-
tional and mathematical methods in medicin, 2015:1–10, 2015. Article ID 303250.
[47] U.B. Mogensen, H. Ishwaran, and T.A. Gerds. Evaluating random forests for survival
analysis using prediction error curves. Journal of Statistical Software, 50(11):1–23,
2012.
[48] J.B. Nasejje, H. Mwambi, K. Dheda, and M. Lesosky. A comparison of the condition-
al inference survival forest model to random survival forests based on a simulation
study as well as on two applications with time-to-event data. BMC Medical Research
Methodology, 17(115):1–17, 2017.
[49] M.Z. Nezhad, N. Sadati, K. Yang, and D. Zhu. A deep active survival analysis ap-
proach for precision treatment recommendations: Application of prostate cancer.
arXiv:1804.03280v1, April 2018.
[50] I.K. Omurlu, M. Ture, and F. Tokatli. The comparisons of random survival forests and
cox regression analysis with simulation and an application related to breast cancer.
Expert Systems with Applications, 36:8582–8588, 2009.
[51] M. Pagano and K. Gauvreau. Principles of biostatistics. Pacific Grove, CA : Duxbury,
2000.
[52] S. Polsterl, N. Navab, and A. Katouzian. An efficient training algorithm for kernel
survival support vector machines. arXiv:1611.07054v, Nov 2016.
[53] R. Ranganath, A. Perotte, N. Elhadad, and D. Blei. Deep survival analysis.
arXiv:1608.02158, September 2016.
[54] C.A. Ronao and S.-B. Cho. Random forests with weighted voting for anomalous query
access detection in relational databases. In Artificial Intelligence and Soft Computing.
ICAISC 2015, volume 9120 of Lecture Notes in Computer Science, pages 36–48, Cham,
2015. Springer.
[55] W. Sauerbrei and P. Royston. Building multivariable prognostic and diagnostic models:
transformation of the predictors by using fractional polynomials. Journal of the Royal
Statistics Society Series A, 162(1):71–94, 1999.
[56] M. Schmid, M.N. Wright, and A. Ziegler. On the use of harrell’s c for clinical risk
prediction via random survival forests. Expert Systems with Applications, 63:450–459,
2016.
[57] B. Scholkopf and A.J. Smola. Learning with Kernels: Support Vector Machines, Reg-
ularization, Optimization, and Beyond. The MIT Press, Cambridge, Massachusetts,
2002.

25
[58] M.R. Segal. Regression trees for censored data. Biometrics, 44(1):35–47, 1988.

[59] P.K. Shivaswamy, W. Chu, and M. Jansche. A support vector approach to censored
targets. In Seventh IEEE International Conference on Data Mining, ICDM 2007, pages
655–660. IEEE, 2007.

[60] X.G. Su and J.J. Fan. Multivariate survival trees: a maximum likelihood approach
based on frailty models. Biometrics, 60(1):93–99, 2004.

[61] J.M.G. Taylor. Random survival forests. Journal of Thoracic Oncology, 6(12):1974–
1975, 2011.

[62] N. Ternes, F. Rotolo, and S. Michiels. Empirical extensions of the lasso penalty to
reduce the false discovery rate in high-dimensional cox regression models. Statistics in
medicine, 35(15):2561–2573, 2016.

[63] R. Tibshirani. The lasso method for variable selection in the cox model. Statistics in
medicine, 16(4):385–395, 1997.

[64] H. Wang and L. Zhou. Random survival forest with space extensions for censored data.
Artificial intelligence in medicine, 79:52–61, 2017.

[65] P. Wang, Y. Li, and C.K. Reddy. Machine learning for survival analysis: A survey.
arXiv:1708.04649, August 2017.

[66] A. Widodo and B.-S. Yang. Machine health prognostics using survival probability and
support vector machine. Expert Systems with Applications, 38(7):8430–8437, 2011.

[67] D.M. Witten and R. Tibshirani. Survival analysis with high-dimensional covariates.
Statistical Methods in Medical Research, 19(1):29–51, 2010.

[68] M.N. Wright, T. Dankowski, and A. Ziegler. Random forests for survival analysis using
maximally selected rank statistics. arXiv:1605.03391v1, May 2016.

[69] M.N. Wright, T. Dankowski, and A. Ziegler. Unbiased split variable selection for ran-
dom survival forests using maximally selected rank statistics. Statistics in Medicine,
36(8):1272–1284, 2017.

[70] L. Wrobel, A. Gudys, and M. Sikora. Learning rule sets from survival data. BMC
Bioinformatics, 18(1):285–297, 2017.

[71] J. Yao, X. Zhu, F. Zhu, and J. Huang. Deep correlational learning for survival predic-
tion from multi-modality data. In Medical Image Computing and Computer–Assisted
Intervention – MICCAI 2017, volume 10434 of Lecture Notes in Computer Science,
pages 406–414. Springer, Cham, 2017.

26
[72] J. Yoon, W.R. Zame, A. Banerjee, M. Cadeiras, A.M. Alaa, and M. van der Schaar.
Personalized survival predictions via trees of predictors: An application to cardiac
transplantation. PLoS ONE, 13(3:e0194985):1–19, 2018.

[73] H. Zhang. Splitting criteria in survival trees. In 10th Workshop on Statistical Modelling,
volume Lecture notes in statistical series, pages 305–314, New York, 1995. Springer.

[74] H.H. Zhang and W. Lu. Adaptive Lasso for Cox’s proportional hazards model.
Biometrika, 94(3):691–703, 2007.

[75] X. Zhu, J. Yao, and J. Huang. Deep convolutional neural network for survival analysis
with pathological images. In 2016 IEEE International Conference on Bioinformatics
and Biomedicine (BIBM), pages 544–547. IEEE, 2016.

27