Hybrid Algorithm in Estimating Underwater Postmortem Interval: Idepis
Noor Maizura Mohamad Noor*, Amirul Harfirie Ahmad Nubli, Rosmayati Mohemad and Zuriana Abu Bakar
Universiti Malaysia Terengganu, Malaysia
Submission: June 09, 2018;Published: June 20, 2018
*Corresponding author: Noor Maizura Mohamad Noor, Universiti Malaysia Terengganu, School of Informatics and Applied Mathematics, 21030, Kuala Nerus, Terengganu, Malaysia, Email: firstname.lastname@example.org
How to cite this article: Noor M M N, Amirul H A N, Rosmayati M, Zuriana A B. Hybrid Algorithm in Estimating Underwater Postmortem Interval: Idepis. J Forensic Sci & Criminal Inves 2018; 9(4): 555768. DOI:10.19080/JFSCI.2018.09.555768.
Crime investigation is a complex task, which involves huge amounts of information and requires many different types of expert knowledge. Estimation of the postmortem interval (PMI) is one of the most important aspects in forensic practice, mainly to assist in solving any crime investigation when the death has not been witnessed. Determination of the PMI is difficult with the progression of time and soft tissue decomposition, especially for underwater death. Underwater death investigation is challenging as the decomposition of body occurs more rapidly due to many contributing factors. This study aims to develop an enhanced decision-based system in estimating underwater PMI, known as iDepis. The main purpose of iDepis is to estimate the PMI underwater condition. Newton’s Law of cooling is usually used to estimate time of death but it is inaccurate to estimate PMI in underwater cases. With the combination of Henssge Nomogram, estimating the time of death could be ascertained by fixing the contributing factors. To make iDepis more reliable, stages of decomposition are integrated with the interval tree algorithm. Therefore, a combination of Newton’s Law of Cooling, Henssge Nomogram, and stages of decomposition will make the system’s decision reliable along with significant estimation of underwater postmortem interval.
Keywords: Corpse; Decision Support System; Postmortem Interval
Over the past few decades, extensive work has been carried out to determine the Postmortem Interval (PMI) from changes in the biochemical constituents of various body fluids, such as blood, cerebrospinal fluid, and vitreous humor, immediately or shortly after death. PMI is known as the estimation of length of elapsed time since a person has died . The goal of PMI is to determine the fact and subsequently identify the truth that can exonerate a suspect or focus suspicion on a suspect . Accurate estimation of the PMI is a crucial and fundamental step in any crime investigation when the death has not been witnessed. However, the determination of PMI is difficult and becomes less accurate when the body has been immersed for a prolonged period of time especially in underwater deaths . According to the statistics obtained from the World Health Organization and the Center for Disease Control and Prevention, the average of fatal drowning in the United States from 2005 to 2014 is 3,536 people annually [4,5].
Generally, to obtain the estimation time of underwater death either in a lake, river, ocean, or even in a bathtub is challenging as the decomposition of the body occurs more rapidly due to many contributing independent environmental factors such as temperature, bacterial content, salinity, and aquatic animal activity [6,7]. It becomes more challenging if the body has deceased somewhere else and then dragged into the water. Forensic science only works for truth by making sure the examination is complete, the test is performed correctly, the data is interpreted in-depth, and the report is written correctly and easily understood by non-scientists . As of now, there is still a lack of accepted objective methods available that allows for reliable PMI estimation because usually PMI estimation is based on forensic expertise in which there is lack of standard procedure to estimate the time of death.
Thus, the objective of this study is to fill this gap by computerizing the PMI estimation whereby the system will generate the time of death by using a hybrid algorithm which is a combination of the Newton’s Law of Cooling, Henssge Nomogram, and stages of decomposition criteria. Therefore, it makes the results more reliable and significant for estimating the time of death. The remainder of this paper is organized as follows. Section 2 discusses the research background related to the estimation of time of death. Meanwhile, Section 3 explains the method that has been used to estimate the time of death in underwater cases. In Section 4, an architectural design and prototype of the system is presented. Subsequently, Section 5 discusses the experimental setup and result, where a comparison is made of results from the estimation of time of death using rabbit carcasses and the Decision Support System for Underwater Postmortem Interval System (iDepis) and finally, section 6 summarizes with a discussion and the conclusion.
The following sub-sections provide a detailed background
of this study including the Decision Support System (DSS),
Postmortem Interval (PMI), fresh body, Algor Mortis, Rigor
Mortis, Livor Mortis, and Adipocere.
Asante had proposed a combination of algorithms between
Newton’s Law of Cooling and Henssge Nomogram in estimating
the time of death . and the algorithm was improved by
Muhammad . However, the algorithm is not efficient
in estimating the time of death in underwater cases due to
changes in the contributing factors. It is because these factors
cannot be controlled by humans such as the flowing of water,
movement of surrounding air, humidity, and insect activity. The
algorithm could be improved by integrating it with the stages
of decomposition criteria generated from the interval tree
algorithm. Stages of decomposition are mainly used by forensic
experts to estimate the time of death in an underwater situation.
This method is highly efficient because the physical changes of a
corpse can be seen right away in a crime scene.
The PMI, which is also known as the time since death
estimation, aids forensic scientists in death investigations. In
general, there are three (3) common methods typically used to
predict PMI which are stages of decomposition, insect behavior
and life cycles, and environmental factor . PMI determination
is a crucial procedure in forensic practice. Many factors affect the
onset and the course of the PMI changes. Studies have shown
that the ambient temperature, acidity/alkalinity, decomposition
process, bacterial fermentation, and even oxygen pressure will
influence the estimation of the PMI [11,12]. Among all factors,
the ambient temperature is considered the most influential
factor in PMI estimation. There have been increasing efforts
in forensic practice to establish an accurate and objective
method that can be applied in PMI estimation at the ambient
temperature. Most researchers believe that as the cadaver is at a
continuously variable ambient temperature, PMI evaluation of a
single substance at a single temperature is not suitable for PMI
estimation [12,13]. Multivariate approaches, such as generalized
additive models (GAMs) or support vector machines (SVMs),
allow various indicator substances to be incorporated into the
model, thus improving the prediction of the PMIs.
During this broad stage, at the time of death, the heart stops
and the skin gets tight and gray. The cell starts to die (brain 3-7
minutes; skin up to 24 hours). The body muscle will relax, while
the bladder and bowels will empty from the body.
Algor mortis refers to cooling of the body. In death, a body
no longer generates warmth and begins to cool down. As
such, temperature drop is used to estimate the time of death.
Postmortem body temperature (TPM) declines progressively
until it reaches the ambient temperature (TA). Metabolism
generates heat (regulated to a narrow range). The body cools
at a uniform rate, thus the rate of TPM decrease can be used
to accurately determine the time of death . However, body
temperature is a narrow range, not a fixed temperature. The
formula used is denoted in the methodology section under
formulas 1 and 2.
Rigor Mortis commonly refers to the stiffness after death
[10,12]. Immediately after death, the body is limp due to the
relaxation of the muscles. Muscles begin to stiffen due to the
chemical changes in the muscle tissues. Muscles contract when
myosin and acting stick together. Adenosine triphosphate (ATP)
is needed to detach the two. The muscle will stay locked in
a constant state of contraction without ATP (in death) and it
cannot be separated. It occurs in all muscles simultaneously,
but can sometimes be seen in smaller muscles first such as the
face, jaw, neck, and then the trunk and other extremities. There
are three (3) rules to determine the states of rigor mortis. First,
within 10 hours of death, the muscle starts to stiffen. Second, the
whole body will stiffen within 12 to 18 hours and the stiffness
starts to disappear after about 24 hours. Thus, after 24 hours
from the time of death, rigor mortis cannot be used to estimate
the time of death.
Livor Mortis, which is also known as the discoloration of the
body after death . (Forensic Medicine for Medical Students,
2015), occurs due to the gravitational settling of the blood .
The blood starts to pull at the lower part of the body which
causes the upper body discoloration and paleness. It begins
thirty minutes up to three hours after death; the skin gets purple
and waxy. Lips, fingers, and toenails become pale or turn white.
Hands and feet turn blue. Eyes start to sink into the skull.
Adipocere is a skeleton decomposition. It usually takes 9 to
10 days or faster . Commonly, the corpse is immersed in water
or a pool. The water must be warm to ensure the fat undergoes
the hydrolysis process. Based on the summary of the studies, five
criteria have been adopted for application in the iDepis, which
include fresh body, cooling of the body, and stiffness of muscle,
discoloration, and adipocerous, as denoted in Table 1. In order
to estimate the time of death of a corpse, all the criteria will be
integrated with the interval tree algorithm.
Henssge nomogram as illustrated in Figure 1. Is a method to
fix the error in the estimation of time of death. It is also called
the correcting factor. Henssge introduced two (2) nomograms
which can calculate the estimated time of death with rectal
temperature of up to 25℃ and down to 25℃. The estimated
time of death and the correction factor are obtained by drawing
a straight line from the ambient temperature and cross-lined
with the rectal temperature. Therefore, the correcting factor in
underwater cases has been selected as shown in Table 2. The
Newton’s Law of Cooling was introduced by Sir Isaac Newton
in 1701 The formula was originally used to calculate the rate of
temperature change in the body. Later, forensic experts started
using the formula to estimate the time of death. Therefore, the
formula is used in this study as shown in the following:
= 37 − (℃) + 3 (1)
= 98.6 − (℃) (2)
Where: T = Time of death
Rt = Rectal temperature
The algorithm used to derive the stages of decomposition
criteria is the interval tree using the concept of binary search.
Basically, an interval tree is a tree of data structure that holds
the time intervals. It is efficient in finding all the intervals that
overlap with any interval or point. In this case, the criterion
of interval time has been used to determine the overlap time
among the criteria . as shown in Figure 2. he total hours
of overlapping time will be the total estimation of the time of
death. Samples of the algorithm (Geeks for Geeks, n.d.) and the
java pseudo code are stated as follows.
a. Sort the time interval in increasing order.
b. Push the first interval on to a stack
c. For each interval do the following
I. If the current interval does not overlap with the stack
top, push it.
II. If the current interval overlaps with the stack top and
ending time of the current interval is more than that of the
stack top, update the stack top with the ending time of the
d. The end of the stack contains the merged intervals.
e. Get reliable time of death of merged intervals.
The selection process of overlapping time of death among
the criteria executed in interval tree algorithm also can be
illustrated in Figure 2. All of the selected criteria are arranged
in ascending order. Then, the algorithm detects the overlapping
time of death between the criteria. The worst-case criteria
which are unreliable or not overlap will be removed from the
algorithm. The best-case of overlapping criteria will be selected
and the algorithm will select the time that most frequent overlap
to each other. The full flow process hybrid algorithm execution is
illustrated in the Figure 3.
Figure 4 illustrates the architecture of the Decision Support
System for the Underwater Postmortem Interval System
(iDepis) that has been developed to estimate the time of death.
The engine of this system includes criteria such as condition of
air, condition of water, layer of clothing, condition of clothing,
stiffness of corpse, discoloration of corpse, layer of corpse’s
skins, and so on, used along with the interval tree algorithm.
Next, these methods are used in combination with Newton’s Law
of Cooling and Henssge Nomogram. The estimation of the time
of death will be stored in the MySQL database and the analysis
reports will be generated via the Java Server Page (JSP) Figure
5. Shows the screenshot of iDepis, which illustrates the iDepis
calculator and form. The user can utilize this system by selecting
the corpse that has been registered by the police. Then, the user
has to fill in the rectal temperature field for the Newton’s Law of
Cooling. Next, the user selects all the criteria needed from the
crime scene and hits the submit button to generate the results
and a full report. Meanwhile (Figure 6). Illustrates the yearly
report that can be generated from selecting the current or past
year. The system can generate yearly report by categorizing the
cause of death.
An experiment was conducted on 9 January 2017 to validate
and compare the results of the iDepis with rabbit carcasses via
animal etiquette. A science officer was hired to do an experiment
with the standard animal procedure. Rabbit carcasses were
chosen because the composition of their body is similar to the
human body . thus making the estimated time of death
almost similar to human death. There were 6 samples of rabbit
carcasses. Each of them was immersed in water under different
types of conditions and criteria sets. However, there are still
more criteria that we could not control such as surrounding air,
water temperature, humidity, flow of water (damp and lake), and
disturbance of animals.
Figure 7 illustrates the comparison of results of the
estimated time of death by using the Asante algorithm and the
real underwater time of death. Figure 7 shows the comparison
of results of the estimation of the time of death by using the
Asante algorithm and the real underwater time of death against
the types of rabbits. (Figure 8). illustrates the comparison of
results of the estimated time of death by using iDepis and the
real underwater time of death against the types of rabbits. From
the illustration in Figures 7 & 8, Kampung Putih, Lionhead Grey,
Lionhead White, and Kampung Grey represent the samples and
types of rabbit carcasses. After the results for the time of death
are recorded, the standard deviation is calculated [16,17]. This
calculation is important to measure the dispersion that refers
to a distribution’s extent of stretching between values in a set
of data. The lower the standard deviation, the closer the data
points tend to be the expected value, μ. Conversely, a higher
standard deviation indicates a wider range of values.
For the Asante algorithm, the standard deviation is calculated
is 27.68 with mean (x̅ ) 31.67. Based on the results, the standard
deviation shows the data have spread too much from the original
time of death. Moreover, we can observe a huge gap in Figure
6 between the Asante Algorithm’s generated time of death and
the rabbit carcasses’ time of death [18,19]. Next, the data set
of rabbit carcasses and iDepis is used to calculate the standard
deviation. The standard deviation for this sample data is 10.31.
Based on the results of this experiment, the value of the standard
deviation is not spread too much away from the mean (x̅ ), which
is 53.75. Therefore, the data from the iDepis result is reliable for
estimating the time of death for underwater cases.
Decision making on estimating time of death is important
to determine the exact time the corpse deceases. The iDepis
could assist the forensic analyst to obtain the estimation of the
time of death efficiently in a short time. Decision makers in the
forensic unit can calculate the estimation of the time of death
based on a combination of algorithms such as the interval tree,
Henssge Nomogram, and Newton’s Law of Cooling. Certain
information about the corpse can be accessed without wasting
a bundle of paper to store the statistics of the death as it can
be automatically searched through the system by using the web
browser. In the future, in order to improve decision-making, an
artificial intelligent agent needs to be integrated with the iDepis
to assist crime investigation. Furthermore, added features such
as the back-tracking technique could be integrated with the
artificial intelligent agent to improve the estimation of the time
of death. To accomplish these tasks efficiently, a new framework
or model should be designed, the algorithm should be improved,
and the real data from the crime scenes should be employed.
Finally, forensic experts must be involved in validating and
verifying the system.
The authors would like to thank the Royal Police Malaysia
(RPM) for its continuous support. This work is supported by a
grant from the Fundamental Research Grant Scheme (FRGS) and
the Ministry of Higher Education (MOHE) with the vot number