Bayesian Inference for a Time-Fractional HIV Model with Nonlinear Diffusion
Mohamed BenSalah, Salih Tatar, Süleyman Ulusoy · 2025
Read Bayesian Inference for a Time-Fractional HIV Model with Nonlinear Diffusion by Mohamed BenSalah, Salih Tatar, Süleyman Ulusoy (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-23638.txt.
Introduction
Mohamed BenSalah and co-authors tackle a hard inference problem in the study of HIV spread. Their 2025 paper uses a model that tracks healthy target cells, infected cells, and free virus over time. The model has a fractional time term and spread terms that are not fixed in advance. The goal is to recover those unknown spread functions from data taken at one final time.
Such inverse problems are unstable, and noisy data make them worse. The team therefore uses Bayesian inference, which gives estimates along with a measure of doubt. Their method, an iterative ensemble Kalman approach, needs no gradient data and can fit several spread terms at once. Tests under varied noise levels show the method rebuilds the true functions well.
Catalog record
| CREATOR | Mohamed BenSalah, Salih Tatar, Süleyman Ulusoy |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-23638
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Bayesian Inference for a Time-Fractional HIV Model with Nonlinear Diffusion Mohamed BenSalah, Salih Tatar, Süleyman Ulusoy Abstract This study investigates an inverse problem associated with a time-fractional HIV infection model incorporating nonlinear diffusion. The model describes the dynamics of uninfected target cells, infected cells, and free virus particles, where the diffusion terms are nonlinear density functions. The primary objective is to recover the unknown diffusion functions by utilizing final-time measurement data. Due to the inherent ill-posedness of the inverse problem and the presence of measurement noise, we employ a Bayesian inference framework to obtain stable and reliable estimates while quantifying uncertainty. To solve the inverse problem efficiently, we develop an Iterative Regularizing Ensemble Kalman Method (IREKM), which enables the simultaneous estimation of multiple diffusion terms without requiring gradient information. Numerical experiments validate the effectiveness of the proposed method in reconstructing the unknown diffusion terms under different noise levels, demonstrating its robustness and accuracy. These findings contribute to a deeper understanding of HIV infection dynamics and provide a computational approach for parameter estimation in fractional diffusion models. Keywords: HIV model, time-fractional derivative, Nonlinear density function, Inverse problem, Bayesian inference; ensemble Kalman method AMS Subject Classifications: 26A33, 35R11, 35R30, 62F15, 92D25 1 Introduction In this paper, we study an inverse problem for the following time-fractional HIV infection model: \displaystyle\displaystyle\left\{\begin{array}[]{l}\partial_{t}^{\alpha}u_{1}-\text{div}(D_{1}(u_{1})\nabla u_{1})=-du_{1}-\beta(u_{1},u_{3})-\sigma(u_{1},u_{2}),(x,t)\in\Omega_{T},\\ \partial_{t}^{\alpha}u_{2}-\text{div}(D_{2}(u_{2})\nabla u_{2})=\beta(u_{1},u_{3})+\sigma(u_{1},u_{2})-ru_{2},\quad(x,t)\in\Omega_{T},\\ \partial_{t}^{\alpha}u_{3}-\text{div}(D_{3}(u_{3})\nabla u_{3})=Nu_{2}-eu_{3},\qquad\qquad\qquad\qquad(x,t)\in\Omega_{T},\\ u_{1}(x,t)=u_{2}(x,t)=u_{3}(x,t)=0,\penalty\ \qquad\qquad\qquad\qquad\qquad(x,t)\in\Gamma_{T},\\ u_{1}(x,0)=u_{1,0}(x),u_{2}(x,0)=u_{2,0}(x),u_{3}(x,0)=u_{3,0}(x),\qquad x\in\Omega,\end{array}\right. where \displaystyle\partial_{t}^{\alpha} is the Caputo fractional derivative of order \displaystyle 0<\alpha<1 , \displaystyle\Omega\subset\mathbb{R}^{n} is a bounded domain with a boundary \displaystyle\partial\Omega , \displaystyle T>0 is a final time, \displaystyle\Omega_{T}=\Omega\times(0,T) , \displaystyle\Gamma_{T}=\partial\Omega\times(0,T) , \displaystyle u_{1} is the concentration of healthy cells, \displaystyle u_{2} is the concentration of infected cells, \displaystyle u_{3} is the concentration of healthy cells virions, \displaystyle D_{1}(u_{1}),D_{2}(u_{2}) and \displaystyle D_{3}(u_{3}) are the nonlinear density functions for \displaystyle u_{1},u_{2} and \displaystyle u_{3} respectively, \displaystyle d=d(x,t)>0 is the death rate for healthy cells, \displaystyle N=N(x,t)>0 is the virus production rate, \displaystyle e=e(x,t)>0 is the death rate of virions, \displaystyle r=r(x,t)>0 is the death rate of infected cells, \displaystyle\beta(u_{1},u_{3})=\beta_{1}\,\frac{u_{1}u_{3}}{1+u_{1}} , \displaystyle\sigma(u_{1},u_{2})=\beta_{2}\,\frac{u_{1}u_{2}}{1+u_{1}} , \displaystyle\beta_{1}=\beta_{1}(x,t)>0 is the virus infection rate, \displaystyle\beta_{2}=\beta_{2}(x,t)>0 is the cell-cell infection rate. We assume that \displaystyle u_{1,0}(x),u_{2,0}(x) and \displaystyle u_{3,0}(x) are in \displaystyle L_{2}(\Omega) , the set of square-integrable functions in \displaystyle\Omega and \displaystyle d(x,t),\beta_{1}(x,t),\beta_{2}(x,t),r(x,t),N(x,t),e(x,t) are in \displaystyle C(\Omega_{T}) , the set of continuous functions on \displaystyle\Omega_{T} . For the given inputs \displaystyle\alpha , \displaystyle T , \displaystyle D_{1}(u_{1}),D_{2}(u_{2}),D_{3}(u_{3}) , \displaystyle d , \displaystyle\beta , \displaystyle\sigma , \displaystyle r , \displaystyle e , \displaystyle N , \displaystyle u_{1,0} , \displaystyle u_{2,0} , \displaystyle u_{3,0} , the problem ( 1 ) is called the direct problem. HIV (human immunodeficiency virus) attacks immune cells, impairing the body’s ability to combat infections and making individuals more susceptible to opportunistic diseases. If left untreated, HIV can progress to AIDS (acquired immunodeficiency syndrome). In recent years, several mathematical models have been developed to describe the dynamics of HIV infection and to analyze epidemic patterns associated with AIDS. For instance, in [ 1 ] , a reaction-diffusion within-host HIV model incorporating cell mobility, spatial heterogeneity, and cell-to-cell transmission is proposed. The authors investigate the well-posedness and global dynamics of the model and analyze the influence of spatial heterogeneity and diffusion properties on virion propagation through numerical simulations. In [ 2 ] , the authors introduce a mathematical model to examine the interaction between drug addiction and the spread of HIV/AIDS in Iranian prisons, studying the stability of the system. A separate study [ 3 ] formulates an ordinary differential equation model describing the dynamics of HIV/AIDS, focusing on control strategies and determining optimal conditions for disease management. Furthermore, [ 4 ] presents a preliminary analysis of HIV transmission dynamics, leveraging quantitative epidemiological data and simple deterministic models to describe viral spread and persistence. In [ 5 ] , a mathematical model of HIV/AIDS transmission through sexual contact is proposed, incorporating asymptomatic and symptomatic phases as a system of differential equations. An epidemic model with multiple stages of infection and its global stability is analyzed in [ 6 ] . Another study [ 7 ] proposes an HIV dynamics model under distinct HAART (highly active antiretroviral therapy) regimens, predicting disease progression across all infection stages. However, fractional differential equations offer significant advantages over integer-order models in capturing the behavior of dynamical systems (see [ 8 ] - [ 17 ] and references therein). Based on this premise, the authors in [ 18 ] modified the model from [ 1 ] to study HIV spread and transmission dynamics within the host. The modifications include introducing time-fractional derivatives, density-dependent diffusion operators, and Holling type II functional responses to better capture HIV infection mechanisms within healthy cells. It is more realistic to consider density-dependent diffusion rather than linear diffusion in biological models, as the movement of infected cells through normal tissue resembles diffusion in a porous medium. Consequently, the random motility of infected cells is best represented as a function of the system’s state variables. Recently, there has been growing interest in inverse problems involving fractional derivatives. In most of these studies, a fractional time derivative is considered, and the inverse problem typically involves determining this derivative, a source term, or a coefficient under some additional conditions. These problems are of significant physical and practical importance. However, the majority of existing research focuses on linear models (see [ 19 ] - [ 31 ] and the references therein). Inverse problems for nonlinear and nonlocal models constitute a relatively new area of study, with only a limited number of experimental investigations and analytical results. For instance, in [ 32 ] , a nonlinear time-fractional inverse coefficient problem is examined, where the unknown coefficient depends on the gradient of the solution. The authors prove the existence of a unique solution to the direct problem and establish the existence of a quasi-solution for the inverse problem within the class of admissible coefficients. A similar problem, in which the unknown coefficient depends on the solution itself, is studied in [ 33 ] . In [ 34 ] , the authors analyze numerical solutions for both the direct and inverse problems of a nonlinear, nonlocal time-fractional equation. They propose a numerical method based on discretizing the minimization problem and employing the steepest descent method alongside a least-squares approach to solve the inverse problem. The direct and inverse problems for a nonlinear time-fractional diffusion equation are investigated in [ 35 ] , where a quasi-Newton optimization method is applied to obtain the numerical solution of the inverse problem. To address the ill-posedness of the inverse problem, Tikhonov regularization is utilized. In [ 36 ] , an iterative method of lines scheme is developed for the numerical solution of the time-fractional Richards equation with implicit Neumann boundary conditions. Additionally, [ 37 ] examines an inverse problem for an inhomogeneous time-fractional diffusion equation in a one-dimensional real-positive semiaxis domain. The authors demonstrate that the inverse problem is severely ill-posed and apply a modified regularization method based on frequency-domain solutions to address this issue. In the present study, we investigate an inverse problem for the direct problem ( 1 ). Unlike the aforementioned studies, we consider a nonlinear and nonlocal system with nonlinear functions on the side, where three unknown nonlinear density