 Research note
 Open Access
 Published:
Optimal control strategy for the effects of hard water consumption on kidneyrelated diseases
BMC Research Notes volume 13, Article number: 201 (2020)
Abstract
Objectives
We study the optimal control strategy for the effects of hard water consumption on kidneyrelated diseases. The mathematical model has been formulated and studied to gain insights on the optimal control strategy on the effects of hardwater consumption on kidneyrelated diseases. The positivity and boundedness of the solutions are determined. A global sensitivity analysis has been performed and the numerical solutions have been carried out.
Results
A global sensitivity analysis shows that the control on water is an important parameter. This can reduce the proportion of individuals with kidneydysfunction and hence reduces the proportion of individuals with kidneyrelated diseases. Furthermore, the numerical solutions show that with the optimal control, the proportion of individuals with kidneyrelated diseases can be minimised.
Introduction
Water is an essential need of human. The increasing population and industrialization result in the contamination of water which can be a source of infectious and noninfectious diseases. Waterborne diseases are one of the major problems in the world and contribute to 2 million deaths annually [1].
Water is a pure substance consisting of hydrogen and oxygen. However, water is rarely found in its pure form. It contains the other ions. Hardwater is the water containing higher level of calcium and magnesium [2]. A longterm consumption of hard water leads to the kidney dysfunction which can cause the otherrelated diseases [2, 3]. As its vital function is to excrete toxins and the overload amount of any substances in the human body [4, 5], the dysfunction kidneys can be considered as a silent killer [3] and is a major problem in most developing countries [6]. In Nusa Tenggara Timur (NTT), Indonesia, the water sources, which are generally consumed by the majority of citizens, mostly contain higher concentrations of calcium and magnesium ions [7]. This might be one of the underlying factors contributing to a higher number of individuals with kidney problems [8]. Data showed that NTT is in the top four province in Indonesia with a higher number of individuals with kidney problems [8].
Mathematical models are commonly used to understand the complex phenomena [9,10,11]. Although a number of mathematical models have been developed to understand the transmission of waterborne diseases such as cholera and typhoid [12, 13], only small number of mathematical models of kidneyrelated diseases have been developed [14,15,16,17,18]. Tambaru et al. [7] formulated a mathematical model for kidney dysfunction and found that the control on water can reduce the number of individuals with kidney dysfunction. The model includes a control parameter on water only. Furthermore, the control parameter did not depend on time. Ndii et al. [14, 15] formulated a mathematical model for the effects of hard water consumption on kidney dysfunctions and kidneyrelated diseases and formulated a numerical scheme to solve the model. However, they did not include the effects of treatments/controls on the model. Walk et al. [16] developed a model to predict how myeloma cells collectively behave and proteins that involved in kidney damage. This research focuses on the interaction between cells in the proximal tubule of the kidney, free light chains, renal fibroblasts, and myeloma cells. Other mathematical models related to kidney diseases were kidney failure diagnostics model using the Artificial Neural Network (ANN) [17] and the effects of proliferative kidney diseases model using the differential equations [18]. Although these studies investigated the kidney problems but the effects of hard water consumption on kidney problems and its controls are not well studied. To the best of our knowledge, little modeling research has been conducted to investigate the effects of the consumption of hard water on the kidneyrelated diseases. Motivated by this, in this paper, we analyse the problem by the use of a mathematical model. In this paper, the effects of control and a global sensitivity analysis have been analysed.
Main text
Methods and results
Formulation of mathematical model
This section presents formulation of a mathematical model of the effects of hard water consumption on kidneyrelated diseases. We extend the model of Tambaru et. al. [7] and Ndii et al. [14] by including the kidneyrelated disease compartment and the control parameters. Let S be susceptible class, I be kidney dysfunction class, \(I_R\) be the kidneyrelated diseases class, and R be recovered class. Individuals experience kidney dysfunction when they consume hard water at a rate \(\beta \lambda (W)\). After a certain period, they attract the kidneyrelated disease at a rate \(\gamma\). Individuals with kidney dysfunctions recover after they undergo the treatment at a rate \(p_1u_1\) where \(p_1\) is the recovery probability due to treatment and \(u_1\) is the treatment/control rate. The individuals with kidney related diseases recover due to selfrecovery at a rate \(\tau\) and treatment/control at a rate \(p_2u_2\) where \(p_2\) is the recovery probability due to treatment and \(u_2\) is the treatment/control rate. An increase level of the hardness of water is determined by the parameter b and limited by the carrying capacity, K. The level of hardness of water decreases when the control on water is implemented at a rate \(u_3\). The model is governed by the following system of differential equations
The \(\lambda (W)\) is the probability of individuals attracting kidney dysfunction which depends on the concentrations of calcium and magnesium in the water which is governed by the following equation,
It is clear that the maximum probability of catching kidney dysfunction is set to be 0.5 and therefore, the maximum concentrations of calcium and magnesium in the water is equal to carrying capacity, K. The A is \(\mu N\) where N is the total human population. We assume a constant human population. The control parameters (\(u_1(t)\), \(u_2(t)\), \(u_3(t)\)) are functions of time. Furthermore, all parameters are positive and the initial conditions are given by: \(S(0)>0\), \(I(0)>0\), \(I_R(0)>0\), \(W(0)>0\).
The parameter descriptions, ranges of values, references, and units are given in Additional file 1: Table S1. This is used in the numerical simulation and sensitivity analysis.
Analysis of the model
Nondimensionalization of the model
In this section, we nondimensionalised the model. To make system (1) dimensionless, we made the following substitution: \(S=sN\), \(I=iN\), \(I_R=i_rN\), \(R=rN\), and \(W=wK\). We obtain the following system of equation
where
Positivity and boundedness of the solution
This section presents the positivity and boundedness of the mathematical model. The model describes the changes in human population and hence it is wellposed if it satisfies the positivity and boundedness conditions.
Theorem 1
Given that the initial conditions of system are positive, the solutionss(t), i(t), \(i_r(t)\)andw(t) are nonnegative for all\(t>0\).
Proof
Assume that \(T=\sup \{t>0, s>0, i>0, i_r>0, w>0 \} \in (0, t]\). Clearly, \(T>0\). From the first equation of the Eq. (3), we obtain
By integrating (5) from 0 to T to obtain
Similarly, it can be shown for \(i(t)>0\), \(i_r(t)>0\), \(r(t)>0\), and \(w(t)>0\) for all \(t>0\). \(\square\)
Theorem 2
Let
Define\(\Omega =\Omega _H \times \Omega _B\). If\(N(0)\le 1\)and\(w(0)\le 1\frac{u_3}{b}\), then the region\(\Omega\)is positively invariant for Model (3) for nonnegative initial conditions.
Proof
Let \(N=s+i+i_r+r\). Then we obtain
Assuming that \(N(0)\le 1\), we conclude that \(N(t)\le 1\). Therefore, the Eq. (6) defines the biologically feasible region for the human population. For the concentrations of calcium and magnesium in the water, it follows that
If \(w(0)\le 1\frac{u_3}{b}\), then \(w(t)\le 1\frac{u_3}{b}\). Therefore, Eq. (7) defines the biologically feasible region for the concentration of calcium and magnesium in the water. From Eqs. (6) and (7), we know that N(t) and w(t) are bounded for all \(t>0\). Therefore, every solution of Model (3) with initial condition in \(\Omega\) remains in \(\Omega\). \(\square\)
Sensitivity analysis
In sensitivity analysis, we use the combination of Latin hypercube sampling (LHS) and Partial rank correlation coefficient (PRCC) multivariate analysis to determine the most influential parameters of the model [19]. First, we measure against the increasing proportion of individuals with kidney dysfunction which is
The results of sensitivity analysis are given in Additional file 1: Figure S1.
Additional file 1: Figure S1 shows that the parameters \(\beta\), b, \(u_3\) are the most influential parameters. The first two have a positive relationship and the last one has a negative relationship.
We measure against the increasing proportion of individuals with kidneyrelated diseases, which is
Figure 1 presents the results of a sensitivity analysis when measured against the increasing proportion of individuals with kidneyrelated diseases. It shows that the parameter \(\gamma\), \(\beta\), \(b\,u_1\) and \(u_3\) are the most influential parameters where the first three have the positive relationship and the latter has negative relationship.
Optimal control analysis
In an optimal control approach, we define the objective functional as follows
The \(\zeta _1\), ..., \(\zeta _6\) are the weight constants for individuals with kidney dysfunction, kidneyrelated diseases, the concentrations of calcium and magnesium in the water, the cost of control on i, the cost of control on \(i_r\), and the cost of control on w, respectively. The necessary conditions that an optimal control have to satisfy are based on the Pontryagin’s Maximum Principle [20]. The Hamiltonian function is the following
Theorem 3
Given optimal controls\((u_1, u_2, u_3)\)and the solutions of the state\((s^*, i^*, i_r^*, w^*)\)which minimises\(J(u_1, u_2, u_3)\)overU. There exist adjoint variables\(\lambda _s\), \(\lambda _i\), \(\lambda _{i_r}\), \(\lambda _r\), \(\lambda _w\)satisfying
with transversality condition\(\lambda _l(t_f)=0\), where\(l=(s, i, i_r, r, w)\). The optimality condition is given by
The controls\((u_1, u_2, u_3)\)are given by
Proof
The differential equations governing the adjoint variables are obtained by differentiating the hamiltonian function in respect to state variables. The adjoint variables are
The transversality condition \(\lambda _l(t_f)=0\) where \(l=s, i, i_r, r, w\). Furthermore, we differentiate the hamiltonian function in respect to control variables to obtain
Using the bounds of the controls, we obtain the characterisation of the control as given in Eq. (12). \(\square\)
Numerical simulation
This section presents numerical simulations of the model. Most parameters are strongly uncertain and hence further research needs to be conducted to obtain the precise values. In the numerical simulation, the following parameter values are used: \(\mu =1/65\), \(\beta =0.1\), \(\gamma =1/5\), \(\tau =1/5\), \(p_1=0.5\), \(p_2=0.5\), \(b=0.05\). The parameter values are taken from literature. When individuals obtain kidney dysfunction, it takes around 1–10 years to progress to kidneyrelated diseases. In our simulation, we use 5 years and hence the parameter \(\gamma =1/5\). The recovery probability of the controls (\(p_1\), \(p_2\)) is assumed to be 0.5. The human lifespan in Nusa Tenggara Timur is around 65 years and hence the human death rate (\(\mu _H\)) is taken to be 1/65. For the weight constant, we use \(\zeta _1=\zeta _2=\zeta _3=\zeta _4=\zeta _5=\zeta _6=1.0\) [21]. The values of weight constants used in the numerical simulation are only of theoretical sense to illustrate the effects control in this paper. Furthermore, the initial proportions of the population are \(S_H(0)=0.9\), \(I_H(0)=0.1\), \(I_R(0)=0\), \(R(0)=0\), \(W(0)=0.1\).
Figure 2 illustrates the numerical solutions of the model with and without control. The proportion of individuals with kidney dysfunction and kidneyrelated diseases decreases by approximately 20% and 17% respectively after the control is implemented. The control profile is given in Fig. 3b. It can be seen that the control rate on water is at highest level and decreases at the end of period.
We vary the weight \(\zeta _5\) associated with the costs of control on individuals with kidneyrelated diseases. We use the value the weights \(\zeta _5=0.1\), \(\zeta _5=1\), \(\zeta _5=10\), \(\zeta _5=100\). The values represent very cheap, cheap, expensive and very expensive costs of controls [21]. The other weights are kept constant at the value of one. The values used in the simulation is theoretical and this is sufficient to investigate our purpose: compare the controls profile with different values of weights. The results are given in Fig. 3.
Figure 3 shows that an increase in the weights \(\zeta _5\) do not significantly affects the control rate except for the \(\zeta =0.1\). When the values of \(\zeta _5\) is one, the control rates \(u_1\) and \(u_2\) are almost similar at year 6 after that the control rate \(u_1\) increases and decreases at the end of the period. We also conducted the variation the weights, \(\zeta _4\) and \(\zeta _6\), but it gives similar results (not shown here).
Discussion and conclusion
A global sensitivity analysis shows that the transmission rate \((\beta )\), an increase rate of the concentrations of calcium and magnesium ions in the water (b), and the control on the water \((u_3)\) are the most influential parameters on the increasing number of individuals with kidney dysfunction. This means that control on water aids in the minimizing the proportion of individuals with kidney dysfunction. Furthermore, the parameter \(\gamma\), \(\beta\), b, \(u_1\) and \(u_3\) are the most influential parameters on the increasing proportion of individuals with kidney dysfunction. This implies that an increase in the control level on individual with kidneydysfunction and water contributes to the reduction in the proportion of individual with kidneyrelated diseases.
Optimal control approach has been conducted to determine the effects of control on disease dynamics. We found that around 9% reduction in the concentration of calcium and magnesium in the water leads to 20% and 17% reduction in the proportion of individuals with kidney dysfunction and kidneyrelated diseases. Although the costs of control associated with kidney dysfunction, kidneyrelated diseases and water increases (very cheap, cheap, expensive and very expensive), the results are similar. That is, the control on water is at high level followed by the control on kidneydysfunction and kidneyrelated diseases. This means that the costs of controls do not significantly affect the level of control. The results are realistic since the process of transmission is on one direction: consumption of hard water resulting in kidney dysfunction and leading to kidneyrelated diseases. Therefore, in order to reduce the proportion of individuals with kidneyrelated diseases, control on water should be implemented. The results are similar to that of sensitivity analysis which shows that the waterrelated parameters are the important parameters.
Limitations
Although the model provides general insights on the effects of hard water consumption on kidney related diseases, it presents a general theoretical results only. Therefore, the model can be extended to study a specific waterrelated diseases due to the consumption of hard water. As the parameter values are strongly uncertain, further research needs to be undertaken to obtain the precise parameter values. These are the subjects of future work.
Availability of data and materials
The other figure and table have been included in the additional document.
Abbreviations
 NTT:

Nusa Tenggara Timur
 LHS:

Latin hypercube sampling
 PRCC:

Partial rank correlation coefficient
References
 1.
World Health Organization. Waterborne disease related to unsafe water and sanitation. https://www.who.int/sustainabledevelopment/housing/healthrisks/waterbornedisease/en/. Accessed 27 Jan 2020.
 2.
Sengupta P. Potential health impacts of hard water. Int J Prev Med. 2013;4:866–75.
 3.
Wasana HMS, Perera GDRK, Gunawardena PDS, Fernando PS, Bandara J. WHO water quality standards Vs synergic effect(s) of fluoride, heavy metals and hardness in drinking water on kidney tissues. Sci Rep. 2017;7:42516.
 4.
Pizzorno J. The kidney dysfunction epidemic, part 1: causes. Integr Med. 2015;14(6):8–13.
 5.
Pizzorno J. The kidney dysfunction epidemic, part 2: intervention. Integr Med. 2016;15(1):8–12.
 6.
Abraham G, Varughese S, Thandavan T, Iyengar A, Fernando E, Naqvi SAJ, Sheriff R, UrRashid H, Gopalakrishnan N, Kafle RK. Chronic kidney disease hotspots in developing countries in South Asia. Clin Kidney J. 2016;9(1):135–41.
 7.
Tambaru D, Djahi BS, Ndii MZ. The effects of hard water consumption on kidney function: insights from mathematical modelling. In: AIP conference proceedings. vol. 1937, No. 1. p. 020020. 2018. https://doi.org/10.1063/1.5026092.
 8.
Badan Penelitian dan Pengembangan Kesehatan Kementrian Kesehatan Republik Indonesia: Riset Kesehatan Dasar 2013. http://www.depkes.go.id/resources/download/general/Hasil. Accessed 31 Jan 2020.
 9.
Ndii MZ, Carnia E, Supriatna AK. Mathematical models for the spread of rumors: a review. In: Gaol FL, Hutagalung F, Chew FP, editors. Issues and trends in interdisciplinary behavior and social science. Boca Raton: CRC Press; 2018. p. 266–90.
 10.
Ndii MZ, Amarti Z, Wiraningsih ED, Supriatna AK. Rabies epidemic model with uncertainty in parameters: crisp and fuzzy approaches. In: IOP conference series: materials science and Engineering. vol. 332. p. 012031. 2018. https://doi.org/10.1088/1757899x/332/1/012031.
 11.
Ndii MZ, Anggriani N, Supriatna AK. Application of differential transformation method for solving dengue transmission mathematical model. In: AIP conference proceedings. vol. 1937, No. 1, p. 020012. 2018. https://doi.org/10.1063/1.5026084.
 12.
Nyabadza F, Aduamah JM, Mushanyu J. Modelling cholera transmission dynamics in the presence of limited resources. BMC Res Notes. 2019;12(1):475. https://doi.org/10.1186/s1310401945049.
 13.
Yang J, Modnak C, Wang J. Dynamical analysis and optimal control simulation for an agestructured cholera transmission model. J Franklin Inst. 2019;356(15):8438–67. https://doi.org/10.1016/j.jfranklin.2019.08.016.
 14.
Ndii MZ, Tambaru D, Djahi BS. The effects of hard water consumption on kidneyrelated diseases. In: AIP conference proceedings. (Accepted)
 15.
Ndii MZ, Tambaru D, Djahi BS. A nonstandard finite difference scheme for waterrelated disease mathematical model. Appl Math Inf Sci. 2019;13(4):545–51.
 16.
Walk JC, Ayati BP, Holstein SA. Modeling the effects of multiple myeloma on kidney function. Sci Rep. 2019;9(1):1726. https://doi.org/10.1038/s41598018381297.
 17.
Almarashi A, Alghamdi M, Mechai I. A new mathematical model for diagnosing chronic diseases (kidney failure) using ann. Cogent Math Stat. 2018;5(1):1559457. https://doi.org/10.1080/23311835.2018.1559457.
 18.
Carraro L, Mari L, Hartikainen H, Strepparava N, Wahli T, Jokela J, Gatto M, Rinaldo A, Bertuzzo E. An epidemiological model for proliferative kidney disease in salmonid populations. Parasit Vectors. 2016;9(1):487. https://doi.org/10.1186/s130710161759z.
 19.
Marino S, Hogue IB, Ray CJ, Kirschner DE. A methodology for performing global uncertainty and sensitivity analysis in systems biology. J Theor Biol. 2008;254(1):178–96. https://doi.org/10.1016/j.jtbi.2008.04.011.
 20.
Pontryagin LS. Mathematical theory of optimal processes. Classics of Soviet Mathematics. Taylor & Francis. 1987. https://books.google.co.id/books?id=kwzq0F4cBVAC.
 21.
Agusto FB, Leite MCA. Optimal control and costeffective analysis of the 2017 meningitis outbreak in nigeria. Infect Dis Modell. 2019;4:161–87. https://doi.org/10.1016/j.idm.2019.05.003.
Acknowledgements
MZN, DT, and BSD thank the Ministry of Research, Technology and Higher Education (currently Ministry of Education and Culture) for the research grant under the Penelitian Dasar Unggulan Perguruan Tinggi Scheme 2019–2020 (Grant No: 50/UN15.19/PL/2019). We thank the reviewer for suggestions which have greatly improved the manuscript.
Funding
This research has been funded by Indonesian Government through Penelitian Dasar Unggulan Perguruan Tinggi 20192020 (Grant No: 50/UN15.19/PL/2019).
Author information
Affiliations
Contributions
MZN designed the research, formulated the model and carried out the sensitivity analysis, and wrote the manuscript. FB participated in the formulation of the model and performed numerical simulation. DT helped in the designed the research, and wrote the manuscript. ML participated in the formulation of the model. AR carried out the proof of positivity and boundedness. BSD wrote the code and performed numerical simulations. All authors read and approved the final manuscript.
Corresponding author
Correspondence to Meksianis Z. Ndii.
Ethics declarations
Ethics approval and consent to participate
No ethical approval was required for this project as this is secondary research.
Consent to publish
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Ndii, M.Z., Berkanis, F.R., Tambaru, D. et al. Optimal control strategy for the effects of hard water consumption on kidneyrelated diseases. BMC Res Notes 13, 201 (2020). https://doi.org/10.1186/s1310402005043z
Received:
Accepted:
Published:
Keywords
 Hard water
 Mathematical model
 Optimal control
 Sensitivity analysis
 Kidney