Mathematical Medicine and Biology: A Journal of the IMA

Cell shape and mechanics are intricately connected and tightly regulated by mechanochemical events including biochemical signaling, cytoskeletal remodeling, and plasma membrane mechanics. While experimental advances in microscopy have shed light on the intricate coordination involved in cell shape change in response to different cues, the ability to conduct three-dimensional simulations in realistic geometries remains an open computational challenge. In this work, we develop a finite-element framework that incorporates advection-diffusion-reaction equations coupled with equations governing the kinematics of a deformable interface representing the cell membrane. We applied this framework to three distinct coupled mechanochemical systems, each governed by geometric partial differential equations, resulting in large deformations of the interface. In all three examples, our simulations revealed the emergence of feedback between cellular signaling, cytoskeletal organization, and cell shape. In our first two sets of simulations, we observed that cell migration and neutrophil protrusion were regulated by membrane tension-mediated feedback. In our final application, we predicted shape changes of a dendritic spine starting from a realistic geometry, and found that the complex shape of the spine gives rise to localized regimes of actin cytoskeleton remodeling not previously observed with idealized geometries. Thus, our finite-element framework allows us to generate new mechanistic insights for biophysical problems.

Neutrophils move through narrow pores, convoluted channels, and tight spaces in tissue to find infection sites. Their ability to sense weak chemical gradients, undergo directed motion, and solve such path-finding problems rests on internal GTPase signaling circuits that control the front protrusion and rear retraction of the cell. Here we explore several variants of known core polarity circuits, with local and long-ranged negative feedback, including inhibitor downstream of Rac, Rac-Rho antagonism, and effects of membrane tension. The resulting reaction-diffusion (RD) equations for Rac and Rho are then used to simulate protrusion-retractions along the edge of a simulated motile cell. We visualize how cells navigate through narrow tracks with sharp corners and weak chemical gradients in 2D. Our metrics for cell performance include polarity initiation, wall-collision intensity, and track completion. In this way, we expose how Rac and Rho, together with their immediate down and upstream components can fine-tune neutrophil motility through complex environments. Author SummaryWhite blood cells, attracted to sites of infection, migrate through complex tissues to find their target. Such movement requires a balance between robust polarity in one direction versus flexibility in response to spatial cues such as obstacles and sharp turns. Here we use mathematical modeling to explore known intracellular circuits that regulate front protrusion and rear retraction in directed cell migration. We test several such circuits in simulations of cells moving along zigzag tracks with sharp turns. We demonstrate that a basic cell polarity circuit, on its own, has limited success, since cells tend to get trapped in sharp corners. Known modulators of this core, which add local negative feedback, mutual front-back antagonism, and long-range feedback from membrane tension, improve cell performance. A cell with the full front-back-membrane tension regulatory circuit avoids delays due to traps and obstacle collisions, and moves swiftly through a convoluted passage to its target site.

Reaction diffusion (RD) systems play a fundamental role in numerous biochemical and biophysical processes. Here, we present a novel analytical framework for solving RD equations by applying the Wentzel Kramers Brillouin Jeffreys (WKBJ) formalism to Ca nanodomains generated by individual membrane channels, a widely used paradigm for intracellular Ca signaling. Previous models have primarily focused on stationary Ca nanodomains while neglecting diffusion and saturation of intracellular Ca buffers and sensors. In contrast, we derive analytical solutions without these simplifying assumptions. Our analysis demonstrates that sustained Ca influx generates continuously expanding distributions of free Ca, whereas Ca bound buffers and sensors propagate as traveling waves. These predictions are supported experimentally by measurements of one-dimensional fluorescence profiles produced by single-channel activity and two-dimensional profiles generated by whole cell Ca currents. The analytical framework developed here readily extends Michaelis Menten type kinetics to reaction diffusion systems and may therefore be broadly applicable to biochemical and biophysical processes in which diffusion cannot be neglected.

Diffuse gliomas are deadly because the individual tumor cells invade - they travel far from the imageable mass, so it is impossible to remove the tumor completely. On the cellular level, glioma cells seem to be in either a "go" state (in which they do not divide) or a "grow" state (in which they do not migrate). We investigate what this tiny choice has to say about the large-scale speed of the invasion front and whether the implication is sufficiently strong to rule out the classical description of the Fisher-Kolmogorov-Petrovsky-Piskunov (Fisher-KPP) type, in which a single phenotype migrates and proliferates. We derive a two-phenotype reaction-diffusion model with density-dependent switching, and we prove the cooperative (quasi-monotone) structure and the associated comparison principle and study travelling-wave solutions of the model. A leading-edge linearization gives minimal front speed as minimizer of an explicit dispersion relation, and direct simulation verifies the predicted speed. In the experimentally relevant fast switching limit, we find a closed-form expression for the speed, that is, we obtain an effective Fisher-KPP equation with rescaled diffusivity and growth rate, with the fractions of the phenotypes. The "go-or-grow" (GoG) front can move at a maximum speed of half the Fisher speed for the same single-cell motility $D$ and proliferation rate $r$, which occurs only when the cells divide their time equally between the two phenotypes. This bound is directly testable: measurement of the front speed, plus independent determination of $D$ and $r$, discriminates the two hypotheses, and in the GoG case, yields recovery of the phenotype balance. We then extend the result to anisotropic (DTI-informed) invasion along white-matter tracts and discuss implications for understanding clinical measurements of growth rate.

Insecticide-treated bed nets (ITNs) are central to malaria control. They serve as physical barriers and chemical agents that deter and kill mosquitoes, thereby reducing transmission; however, this form of protection reshapes the immunology of malaria by reducing exposure to Plasmodium parasites and weakening the development of naturally acquired immunity. Against this background, the present study develops a modeling framework to investigate how this tension between protection and immunity alters malaria dynamics once vaccination is introduced as a complementary control strategy and the optimum combination of ITN and vaccine coverage required for malaria control. Unlike standard models that use a fixed proportional reduction in transmission, this study models ITN coverage and efficacy as coupled, time-dependent processes and immunity driven by exposure to infection and vaccination. Rigorous analysis of the model identifies existence conditions for equilibria and shows that malaria can be contained through the synergistic interaction of vaccination, vector control, and immunity-mediated host dynamics. Parameter values of the model are estimated by fitting the model to confirmed malaria case data and the estimated baseline reproduction number using these parameter values is 1.41 (95% confidence interval: 1.34-1.48), confirming sustained transmission. Simulations of the model show that, although ITNs reduce immunity acquisition, their net effect is to reduce infections and improve recovery and survival. Hence, population-level benefits of ITNs outweigh their immunity-reducing effects (particularly when combined with vaccination), leading to a reduced malaria burden. Comprehensive sensitivity analysis indicates that malaria burden is driven mostly by mosquito biting intensity, population capacity, and transmission probabilities; while mosquito mortality, effective treatment, ITN performance, and vaccine efficacy cause the most significant reductions. Additionally, malaria is uncontrollable with universal ITN use and vaccination at baseline efficacy. While individual interventions can achieve control under low transmission, neither 100% ITN coverage nor 100% vaccine coverage can achieve control under moderate-to-high transmission. However, a strong synergy between ITNs and vaccination allows combinations of high efficacy to achieve containment at realistic coverage levels, suggesting that integrated malaria control involving effective vector control, vaccination, and prompt treatment is needed.

Over-activation of mitogen-activated protein kinase (MAPK) signaling underlies numerous inflammatory pathologies that are treated using synthetic glucocorticoids to activate glucocorticoid receptors (GR) and induce expression of dual-specificity phosphatase 1 (DUSP1), which encodes for MAPK phosphatase 1 (MKP1). Despite its importance, the single-cell het-erogeneity of this spatial and temporal pathway has not been fully quantified, several regulatory mechanisms are unclear, and accurate quantitative predictions are not possible based on existing models. To address this challenge, we combined immunocytochemistry (ICC) and single-molecule inexpensive FISH (smiFISH) to quantify endogenous GR transport and DUSP1 transcription dynamics across thousands of single cells following dexamethasone (Dex) stimulation. Using Chemical Master Equations (CME) and likelihood-based inference, we identified clear mechanisms and reaction rates for Dex-driven GR nuclear import; compartment-specific GR degradation; GR-dependent modulation of DUSP1 promoter activation and transcription burst frequencies; DUSP1 transcription, elongation, and transport; and time-dependent and saturation-limited cytoplasmic degradation. Rigorous model comparisons against endogenous, fixed-cell data identifies nuclear GR degradation as the dominant mechanism of receptor clearance, indicates that GR primarily regulates promoter activation, and highlights time-dependent AU-rich element (ARE)-mediated mRNA degradation as a likely mechanism for DUSP1 clearance. With these mechanisms, the fully-parameterized model quantitatively predicts joint distributions of GR translation and decay dynamics, DUSP1 transcription site activity, and nuclear and cytoplasmic mRNA heterogeneity among clonal cells as functions of time and across seven orders of magnitude for Dex induction concentrations. Together, these results show that total DUSP1 mRNA levels emerge from the balance between GR-driven activation and cytoplasmic mRNA decay, with the inferred model quantitatively predicting single-cell distributions across held-out conditions.

Mathematical models of enzyme cycles form the basis of quantifying key features of metabolism and membrane transport. These models are often integrated into more comprehensive models such as whole-cell models to understand emergent behaviours between interacting components. However, it is currently computationally infeasible to simulate the full dynamical behaviour of every enzyme at a network scale. Model reduction is frequently used to improve computational efficiency, but in general, these approaches do not preserve physical and thermodynamic consistency. Here, we outline a general method for simplifying enzyme kinetics models while retaining mass, charge and energy balance. We base our approach on the bond graph, which is a general methodology for modelling biological systems from fundamental physical laws. This approach ensures that key physical constraints are enforced in every model, regardless of their complexity. Our thermodynamic model reduction framework is readily extended to electrogenic transporters through the coupling of chemical and electrical processes. Through the application of our approach to both hypothetical enzyme cycles and real data from the Na+/K+ ATPase, we show that it can rapidly screen for plausible network structures in circumstances where enzyme catalytic mechanisms may not be fully characterised, facilitating biological discovery and drug development.

Serial low-dose computed tomography (LDCT) scans in patients who are diagnosed with lung cancer during screening offer a history of the densities of tumors and the tissues that surround them during carcinogenesis and cancer progression. We built a CT-scan-resolution computational model to explore how variations in lung tissue density impact tumor growth and evolution in non-small cell lung cancer (NSCLC). Our findings indicate that tumors spread more rapidly through denser tissues when they upregulate glycolysis whilst tumors spread more rapidly through sparser tissues when they upregulate angiogenesis. We used data and images from the National Lung Screening Trial to calibrate our model for untreated lung cancer growth in patients and observed consistency with model predictions in low-density environments. SignificanceOur lung lesion model supports prior studies that find tumors tend to evolve toward angiogenic or glycolytic phenotypes. We demonstrate that these evolutionary strategies may be driven by the surrounding normal tissue density and may be observable on imaging.

There is a vast literature on the voltage gating of ion channels, with a fairly large fraction concerned with potassium channels, especially of the KV1 family, including Shaker. Experimental evidence derived from protein structure has been interpreted to give gating mechanisms that largely disregard water. We propose that the K+ ion, in order to pass through the gating region and enter the cavity pore, must be largely dehydrated. Competitive interactions of each single hydration shell water at the gate, with counterions, protein, or other water molecules, can remove one water at a time. There are several such interactions for the ion hydration shell; for the ion to pass through the gating region, there must be enough such interactions to leave the ion with at most two hydrating water molecules, in which case the gate is open. Protein conformational changes are secondary, small, and mostly unimportant. The hypothesis has a second part: protons, previously shown to be candidate carriers of the gating current (Kariev and Green, JPC B, 2019, Membranes, 2022, 2024) are capable of reaching the gate; adding four protons to the gate prevents dehydration, leaving the ion with at least three hydrating water molecules, enough to block passage. Quantum calculations presented here support the dehydration part of the hypothesis; they also mostly support the second part, concerning the protons, but further work will be required to fully confirm this. The hypothesis explains the experimental finding that the P475D mutant is essentially constitutively open, while the P475S mutant, with a wider gate opening, is closed at all relevant potentials; the computations presented here show the mechanism for this in detail, further confirming the first part of the hypothesis, and largely but not completely confirming the second part, concerning protons, while showing where further work is needed. This mechanism can also qualitatively account for flicker noise and fluctuations, and their consequences.

Objective To estimate the potential health and equity impacts of a tobacco free generation (TFG) and T21 policy (increasing the legal age of sale to 21) in Australia, in the context of a complex market including widespread illicit tobacco and e-cigarette product availability. Design A Markov macrosimulation model, parameterised with yearly net movements between legal smoking, illicit smoking, vaping, and dual use states, combined with a proportional multi-state lifetable. Setting The Australian population, modelled as an open cohort for 40-years. Intervention A 'business-as-usual' (BAU) scenario was compared to TFG and T21 policies, with both starting in 2026. Variations to policy impacts were tested under increasing background illicit market enforcement. Main outcome measures The model estimates the health-adjusted life years (HALYs) and deaths over 40 years, under each scenario, with differences across age and socioeconomic status (SES) presented. Results The TFG policy reduced daily smoking prevalence among 15-24-year-olds to 4.6% (95% uncertainty interval [UI] 3.8-5.7%) in 20 years' time, compared to 7.2% under the T21 policy and 7.9% under BAU trends. Vaping was minimally impacted by either policy. The TFG policy resulted in 178,000 (95% UI 87,800-314,000) HALYs being gained over 40 years. The policy impact was largest when accompanied by increased illicit market enforcement, reducing daily smoking among 15-24-year-olds to 1.4% within 20 years. Both policies had greater prevalence and health impacts on more disadvantaged compared to advantaged SES groups. Conclusion A TFG policy is expected to produce long-term benefits for the Australian population but would be most effective in combination with increased enforcement of illicit tobacco and e-cigarette markets. Novel strategies to increase quitting in addition to reducing uptake are needed to improve tobacco-related outcomes in the short to medium term.

Aedes aegypti is a major vector of arboviruses, and belongs to subfamily Culicinae, a diverse group of mosquitoes with homomorphic sex-determining chromosomes. Males are the heterogametic sex with a dominant male-determining locus (M locus). The M locus and its counterpart m locus are embedded in a region of suppressed recombination, with a large portion of this recombination desert showing significant molecular differentiation despite homomorphy. We developed a mathematical framework to examine M-linked genome editors that specifically target the m-chromosome during spermatogenesis, mimicking the naturally occurring sex ratio distorters (SRDs) in Culicinae that produce male-biased meiotic drives. Unlike previous models for species with heteromorphic sex chromosomes (e.g., X and Y), we incorporate features stemming from the homomorphic nature of the Ae. aegypti sex chromosomes such as varied linkage to the M locus, making the degree of super-Mendelian inheritance readily tunable. We evaluated in silico SRDs with a range of M-linkage and editing efficiencies and established the theoretical foundation for developing highly efficient SRDs that outperform several methods of population suppression. These SRDs can be tuned to mitigate impact on a neighboring population. The framework developed here is suitable for exploring SRD-mediated genetic biocontrol of pests with homomorphic sex chromosomes.

Biomolecular condensates are dynamic assemblies whose functions depend on continuous exchange of molecular components with the surrounding environment. While scaffold molecules drive phase separation and condensate architecture, many functional components are clients that are recruited through interactions with the scaffold-rich environment. Despite their prevalence, how client-scaffold interactions shape client exchange dynamics remains poorly understood. Here, we develop a reaction-diffusion model for client exchange in scaffold-driven condensates, in which clients switch between a scaffold-bound state and an unbound state. Bound clients exchange through scaffold-mediated transport, whereas unbound clients diffuse through the pore space of the condensate. Using the fluorescence recovery of fully photobleached condensates as a measure of client exchange, we compare transport through these two pathways with bound-unbound conversion and identify three limiting regimes. In the slow-conversion regime, bound and unbound clients recover through distinct scaffold- and pore-mediated pathways. In the intermediate-conversion regime, recovery of bound clients becomes limited by client unbinding. In the fast-conversion regime, local equilibrium between bound and unbound clients produces an effective single-state recovery. We further propose a unifying description that connects these regimes and quantitatively captures the apparent recovery timescales extracted from numerical simulations across condensate sizes. Our results provide a framework for interpreting component-specific exchange dynamics, and highlight client size, client-scaffold binding, and condensate porosity as key regulators of client turnover in multicomponent condensates.

Chromosome segregation errors in oocyte meiosis are a leading cause of early miscarriage and congenital disorders in mammals and these errors become more prevalent with advanced maternal age. Although the effects of aging on the functions of critical meiotic proteins and cytoskeletal filaments in oocytes are known, the influence of aging on the force generating capabilities of oocyte spindle components remains largely unexplored. Through the integration of a coarse-grained model and in situ experiments, we compare the long-axis mechanical properties of metaphase I (MI) and II (MII) oocyte spindles from reproductively young and old mice. Increased inter-kinetochore distance in aged MII oocytes agree with a model of age-associated cohesion loss, and kinetochore dynamics in these spindles following laser ablation suggest a similar reduction in inter-kinetochore bridge viscosity. Simultaneously, we find that both cohesive and poleward force generators lose stiffness with advanced age in MI spindles. In total, we quantify the extent to which structural spindle components lose their stiffness and viscosity during maternal aging, highlighting the multifaceted impacts of aging on mouse oocyte spindle mechanics. Significance StatementO_LIMaternal aging influences mammalian oocyte spindles in numerous ways, yet the impacts of aging on the balance of collective spindle forces remain poorly understood. C_LIO_LIIntegrating coarse-grained mechanical modeling with in situ measurements of spindle morphology and kinetochore dynamics, we quantify age-associated changes to the viscosities and elastic stiffnesses of oocyte spindle component parts. C_LIO_LIThis work provides both a characterization of the effects of aging on force production in mammalian oocyte spindles and a blueprint for future studies of spindle force generation in complex biological contexts. C_LI

Directed migration along chemical gradients controls immune surveillance, development, and cancer invasion. However, the same chemical cue can produce different responses depending on its concentration and whether cells move alone or in groups. For example, in steep gradients, isolated malignant lymphocyte cells migrate away from the chemoattractant source, whereas clusters of the same cells continue to migrate toward it. Here, combining computational modeling and experimental observations, we show that this reversal is governed by coupled mechanisms acting across molecular, cellular, and collective scales. At the single-cell level, our model predicts that receptor endocytosis generates a feedback that produces a nonmonotonic surface receptor density with increasing chemoattractant concentration. Above a critical concentration that depends on the cell's volume-to-sensing-area ratio, receptor depletion reverses cell polarity and drives chemorepulsion. However, in clusters, cell-cell contacts reduce the membrane area exposed to ligand, increasing the volume-to-sensing-area ratio, thus increasing the critical concentration and preserving chemotaxis. An agent-based model incorporating these mechanisms quantitatively reproduces the sign reversal of the migration index across gradient steepness and cluster size. We show that collective rearrangements further stabilize chemoattraction with exchanges between the cluster rim and core helping remove chemorepulsive cells from the leading edge, keeping their fraction below the threshold required to reverse cluster migration. The model further predicts, and experiments confirm, that increasing ambient ligand concentration while keeping the gradient fixed reduces cluster chemoattraction. Our results identify receptor trafficking, cell geometry, and cluster fluidity as physical determinants of collective directional decision-making, with implications for immune cell homing, tissue morphogenesis, and cancer dissemination.

Long-term care facilities (LTCs) worldwide experienced disproportionately high infection and mortality rates during the COVID-19 pandemic, where essential care limits opportunities for contact segregation. However, empirical contact data remain scarce, limiting our understanding of how individual contact behaviours shape transmission in these settings. In this study, we developed a stochastic network-based transmission model parameterised using real-world self-reported contact data collected from a median-sized UK LTC unit. By incorporating high-resolution observational data that reflect routine care delivery patterns, we quantified how heterogeneity in contact networks influences outbreak dynamics. We found substantial variation in contact behaviour between individuals, resulting in highly heterogeneous transmission outcomes. Outbreak occurrence, timing, final size, and the likelihood of super-spreading events all varied markedly depending on the structure of the underlying contact network and the characteristics of the index case. Individuals with high contact activity were considerably more likely to initiate large outbreaks than those with fewer contacts. For a per-contact transmission probability of 10%, introduction of infection through the most highly connected individuals resulted in a greater than 75% probability of a large outbreak. Our findings indicate that preventing infection introduction through both residents and staff is critical for outbreak control in LTCs. Individuals with high contact activity were consistently associated with a greater probability of initiating large outbreaks, highlighting the importance of accounting for contact heterogeneity when designing surveillance and infection-control measures. More broadly, this study demonstrates the importance of accounting for contact-network heterogeneity when designing infection prevention and control measures in LTC settings, and highlights the value of integrating empirical contact data with transmission modelling to inform evidence-based outbreak preparedness, targeted surveillance, and infection-control strategies in long-term care facilities.

Background. Evidence of the effectiveness and cost-effectiveness of new vaccines that reduce the burden of respiratory syncytial virus (RSV) in older populations is emerging. The reported cost-effectiveness of these vaccination strategies varies substantially across different settings. This study assessed the cost-effectiveness of older adult-targeted RSV vaccination strategies in the Australian context and compared findings with published evaluations. Methods. We developed an individual-based dynamic transmission model of RSV infection, linked to a clinical pathways and cost-effectiveness model. We modelled different adult vaccination strategies for the general population and the Indigenous population, and present incremental cost-effectiveness ratios (ICERs) as cost per quality-adjusted life year gained, from a healthcare system perspective. Deterministic and probabilistic sensitivity analyses explored drivers of cost-effectiveness and sensitivity of findings to uncertainty in parameter estimates. Results. Vaccinating the general population of older adults in Australia was not found to be cost-effective at a dose price of 100 AUD, but was found to be cost-saving for Indigenous adults, given the higher disease burden in this population. Individual drivers of ICERs in our setting were dose price, hospitalisation incidence and mortality, however conclusions about cost-effectiveness were robust to joint parameter uncertainty. Conclusions. The cost-effectiveness of vaccinating adults against RSV depends on many uncertain and context-specific quantities. Strategies that target high risk populations were found to be cost-effective in Australia due to the larger avertable burden.

Statistical prediction models for binary outcomes are becoming increasingly popular. One significant challenge is calibrating these models to suit the characteristics of a target population that is structurally different from the original population. Calibration is especially challenging when there is no training data available from the target population. To address this problem, we propose a novel calibration method, SimCal, which uses synthetic data generated from the model development data in conjunction with marginal statistics from the calibration cohort. We show that expert judgment modeling (EJM) may be used for calibration if cross-sectional data from the target population are available comprising expert judgments about the potential outcome and the covariates. We describe three alternative calibration approaches when calibration data are lacking: similarity-binning averaging (SBA), adaptive calibration of predictions (ACP), and Elkan calibration. In a simulation study, we compare SBA, ACP, Elkan calibration, and SimCal. R code for applying these methods is provided from the re-analysis of data on coronary artery disease. We illustrate all 5 calibration approaches with a real data set for predicting functional outcome after stroke and all approaches but EJM in the re-analysis of the Cleveland Clinic data. None of the approaches performed convincingly well in all situations. SimCal performed well when model parameters were correctly specified. EJM failed on the stroke data. Further research is urgently required for calibration in the absence of calibration data.

Environmental variability can strongly alter coexistence among competing species and their extinction risk, particularly when population dynamics are shaped by behavioral interactions, such as fear. In this work, we develop a novel stochastic differential equation competition model that incorporates both non-consumptive fear effects and environmental variability to investigate how behavioral interactions influence species coexistence under random fluctuations. Our result reveals that environmental stochasticity can drive species to extinction even when the corresponding deterministic system admits coexistence. In particular, under an explicit stability condition on the fear and competition parameters and sufficiently strong averaged noise intensities, we prove that both competing species become extinct exponentially almost surely. Conversely, we derive a stochastic persistence criterion in terms of fear, competition, and noise-induced suppression parameters for the fearful species. We further demonstrate that environmental noise may reverse classical competition-exclusion outcomes, leading to qualitatively different long-term dynamics from those predicted deterministically. These results provide rigorous thresholds separating stochastic extinction from persistence and highlight the critical role of environmental variability in fear-mediated competitive ecosystems. From an applied perspective, these results provide insight into how behavioral interactions and environmental variability influence species survival, with potential applications in ecological management and conservation.

Understanding how genomic surveillance capacity translates into population health outcomes is critical for designing effective pandemic response systems, yet the interaction between operational design and epidemiological dynamics remains insufficiently characterized. We develop an integrated analytical framework that links a whole-genome sequencing (WGS) - based surveillance network with a two - variant epidemiological transmission model to evaluate how surveillance operations influence variant detection, intervention timing, and health outcomes. The framework combines a modified susceptible - exposed - infectious - recovered - susceptible (SEIRS) model with a detailed operational representation of a centralized WGS surveillance network in India, incorporating sample collection, transport, batching, sequencing capacity, and reporting delays. We simulate 54 scenario combinations defined by three sequencing capacity levels, three sampling proportions, three variant emergence timings, and two variant profiles (high severity - high immune escape and low severity - low immune escape). Detection of a novel variant triggers a modeled intervention consisting of isolation of some diagnosed individuals, increased testing rates across disease states, and expanded access to hospitalization. Across simulations, the time from variant emergence to intervention implementation ranged from 73 to 351 days, depending on operational and epidemiological conditions. Increasing sampling proportion reduced detection time only when sequencing capacity was sufficient; under constrained capacity, higher sampling increased congestion and delayed detection. Expanding capacity from low to nominal levels substantially reduced turnaround times, with diminishing returns at higher capacity. Earlier detection consistently improved intervention effectiveness, with deaths averted ranging from 0.06% to 14.49% across scenarios. The cost per life - year saved ranged from INR 9,137 to INR 326,714 across all configurations, remaining below one to three times India ' s GDP per capita, consistent with established cost - effectiveness thresholds. These results demonstrate that the performance of genomic surveillance systems is jointly determined by operational and epidemiological dynamics. Effective surveillance design, therefore, requires coordinated optimization of sampling strategies and sequencing capacity to enable timely intervention and maximize population health benefits.

Optimizing dose and schedule remains a central challenge in oncology drug development, particularly for immunotherapies where fixed dosing regimens often fail to account for patient specific heterogeneity in tumor-immune dynamics. Here, we present a hybrid quantitative systems pharmacology-reinforcement learning-Monte Carlo Tree Search (QSP-RL-MCTS) framework for personalized immunotherapy dosing that formulates dose selection as a sequential decision-making problem. The approach integrates a mechanistic QSP model of prostate cancer immunotherapy, transcriptomics informed virtual patient populations and data driven AI system comprising reinforcement learning and Monte Carlo tree search. Reinforcement learning is used to learn adaptive generalized dosing policies that optimize treatment outcomes across the population, while Monte Carlo Tree Search provides forward-looking evaluation of RL predicted dosing trajectories to refine patient-specific decisions. On benchmarking against fixed dosing regimens of ipilimumab, the remission rate of the proposed model (95.2%) was comparable to the highest fixed dosing regimen of 10 mg/kg per dose while the median total dose (72 mg/kg) of the proposed model designed regimen was comparable to the lowest fixed dosing regimen of 3 mg/kg per dose. The model is generalizable across different dosing protocols and can be extended to predict optimal dose under different therapeutic scenarios. Analysis of the learned dosing trajectories enables stratification of patients into distinct response groups and identifies drug activity rate as the dominant determinant of long-term treatment outcome. These results demonstrate how mechanistically guided artificial intelligence can transform population-level dose optimization into patient-specific, biologically interpretable treatment strategies for precision immuno-oncology.