Disease X epidemic control using a stochastic model and a deterministic approximation: Performance comparison with and without parameter uncertainties.

BACKGROUND: The spread of infectious diseases can be modeled using deterministic or stochastic models. A deterministic approximation of a stochastic model can be appropriate under some conditions, but is unable to capture the discrete nature of populations. We look into the choice of a model from the perspective of decision making. METHOD: We consider an emerging disease (Disease X) in a closed population modeled by a stochastic SIR model or its deterministic approximation. The objective of the decision maker is to minimize the cumulative number of symptomatic infected-days over the course of the epidemic by picking a vaccination policy. We consider four decision making scenarios: based on the stochastic model or the deterministic model, and with or without parameter uncertainty. We also consider different sample sizes for uncertain parameter draws and stochastic model runs. We estimate the average performance of decision making in each scenario and for each sample size. RESULTS: The model used for decision making has an influence on the picked policies. The best achievable performance is obtained with the stochastic model, knowing parameter values, and for a large sample size. For small sample sizes, the deterministic model can outperform the stochastic model due to stochastic effects. Resolving uncertainties may bring more benefit than switching to the stochastic model in our example. CONCLUSION: This article illustrates the interplay between the choice of a type of model, parameter uncertainties, and sample sizes. It points to issues to be considered when optimizing a stochastic model.

Value of information dynamics in Disease X vaccine clinical trials.

BACKGROUND: Solutions have been proposed to accelerate the development and rollout of vaccines against a hypothetical disease with epidemic or pandemic potential called Disease X. This may involve resolving uncertainties regarding the disease and the new vaccine. However the value for public health of collecting this information will depend on the time needed to perform research, but also on the time needed to produce vaccine doses. We explore this interplay, and its effect on the decision on whether or not to perform research. METHOD: We simulate numerically the emergence and transmission of a disease in a population using a susceptible-infected-recovered (SIR) compartmental model with vaccination. Uncertainties regarding the disease and the vaccine are represented by parameter prior distributions. We vary the date at which vaccine doses are available, and the date at which information about parameters becomes available. We use the expected value of perfect information (EVPI) and the expected value of partially perfect information (EVPPI) to measure the value of information. RESULTS: As expected, information has less or no value if it comes too late, or (equivalently) if it can only be used too late. However we also find non trivial dynamics for shorter durations of vaccine development. In this parameter area, it can be optimal to implement vaccination without waiting for information depending on the respective durations of dose production and of clinical research. CONCLUSION: We illustrate the value of information dynamics in a Disease X outbreak scenario, and present a general approach to properly take into account uncertainties and transmission dynamics when planning clinical research in this scenario. Our method is based on numerical simulation and allows us to highlight non trivial effects that cannot otherwise be investigated.

Optimal Epidemic Control under Uncertainty: Tradeoffs between Information Collection and Other Actions.

BACKGROUND: Recent epidemics and measures taken to control them-through vaccination or other actions-have highlighted the role and importance of uncertainty in public health. There is generally a tradeoff between information collection and other uses of resources. Whether this tradeoff is solved explicitly or implicitly, the concept of value of information is central to inform policy makers in an uncertain environment. METHOD: We use a deterministic SIR (susceptible, infectious, recovered) disease emergence and transmission model with vaccination that can be administered as 1 or 2 doses. The disease parameters and vaccine characteristics are uncertain. We study the tradeoffs between information acquisition and 2 other measures: bringing vaccination forward and acquiring more vaccine doses. To do this, we quantify the expected value of perfect information (EVPI) under different constraints faced by public health authorities (i.e., the time of the vaccination campaign implementation and the number of vaccine doses available). RESULTS: We discuss the appropriateness of different responses under uncertainty. We show that, in some cases, vaccinating later or with less vaccine doses but more information about the epidemic, and the efficacy of control strategies may bring better results than vaccinating earlier or with more doses and less information, respectively. CONCLUSION: In the present methodological article, we show in an abstract setting how clearly defining and treating the tradeoff between information acquisition and the relaxation of constraints can improve public health decision making. HIGHLIGHTS: Uncertainties can seriously hinder epidemic control, but resolving them is costly. Thus, there are tradeoffs between information collection and alternative uses of resources.We use a generic SIR model with vaccination and a value-of-information framework to explore these tradeoffs.We show in which cases vaccinating later with more information about the epidemic and the efficacy of control measures may be better-or not-than vaccinating earlier with less information.We show in which cases vaccinating with fewer vaccine doses and more information about the epidemic and the efficacy of control measures may be better-or not-than vaccinating with more doses and less information.

Hospital-wide surveillance-based antimicrobial treatments: A Monte-Carlo look-ahead method.

BACKGROUND AND OBJECTIVES: We present a heuristic solution method to the problem of choosing hospital-wide antimicrobial treatments that minimize the cumulative infected patient-days in the long run in a health care facility. METHODS: Our solution method is a rollout algorithm. We rely on the stochastic version of a compartmental model to describe the spread of an infecting organism in the health care facility and the emergence and spread of resistance to two drugs. We assume that the parameters of the model are known. Treatments are chosen at the beginning of each period based on the count of patients with each health status, and on stochastic simulations of the future emergence and spread of antimicrobial resistance. The same treatment is then administered to all patients, including uninfected patients, during the period and cannot be adjusted until the next period. RESULTS: In our simulations, our algorithm allows to reduce the average cumulative infected patient-days over two years by 47.0% compared to the best standard therapy, and by 32.2% compared to a similar heuristic algorithm not using surveillance data (significantly at the 95% threshold). CONCLUSION: Our heuristic solution method is simple yet flexible. We explain how it can be used either to perform online optimization, or to produce data for quantitative analysis. Its performance is illustrated using a relatively simple infectious disease transmission model, but it is compatible with more advanced epidemiological models.

Optimal dynamic empirical therapy in a health care facility: A Monte-Carlo look-ahead method.

BACKGROUND AND OBJECTIVES: Empirical antimicrobial prescription strategies have been proposed to counteract the selection of resistant pathogenic strains. The respective merits of such strategies have been debated. Rather than comparing a finite number of policies, we take an optimization approach and propose a solution to the problem of finding an empirical therapy policy in a health care facility that minimizes the cumulative infected patient-days over a given time horizon. METHODS: We assume that the parameters of the model are known and that when the policy is implemented, all patients receive the same treatment at a given time. We model the emergence and spread of antimicrobial resistance at the population level with the stochastic version of a compartmental model. The model features two drugs and the possibility of double resistance. Our solution method is a rollout algorithm. RESULTS: In our example, the optimal policy computed with this method allows to reduce the average cumulative infected patient-days over two years by 22% compared to the best standard therapy. Considering regularity constraints, we could derive a policy with a fixed period and a performance close to that of the optimal policy. The average cumulative infected patient-days over two years obtained with the optimal policy is 6% lower (significantly at the 95% threshold) than that obtained with the fixed period policy. CONCLUSION: Our results illustrate the performance of a highly flexible solution method that will contribute to the development of implementable empirical therapy policies.

Informed and uninformed empirical therapy policies.

We argue that a proper distinction must be made between informed and uninformed decision making when setting empirical therapy policies, as this allows one to estimate the value of gathering more information about the pathogens and their transmission and thus to set research priorities. We rely on the stochastic version of a compartmental model to describe the spread of an infecting organism in a health care facility and the emergence and spread of resistance to two drugs. We focus on information and uncertainty regarding the parameters of this model. We consider a family of adaptive empirical therapy policies. In the uninformed setting, the best adaptive policy allowsone to reduce the average cumulative infected patient days over 2 years by 39.3% (95% confidence interval (CI), 30.3-48.1%) compared to the combination therapy. Choosing empirical therapy policies while knowing the exact parameter values allows one to further decrease the cumulative infected patient days by 3.9% (95% CI, 2.1-5.8%) on average. In our setting, the benefit of perfect information might be offset by increased drug consumption.

Cost effectiveness and policy announcement: The case of measles mandatory vaccination.

In a vaccination game, individuals respond to an epidemic by engaging in preventive behaviors that, in turn, influence the course of the epidemic. Such feedback loops need to be considered in the cost effectiveness evaluations of public health policies. We elaborate on the example of mandatory measles vaccination and the role of its anticipation. Our framework is a SIR compartmental model with fully rational forward looking agents who can therefore anticipate on the effects of the mandatory vaccination policy. Before vaccination becomes mandatory, parents decide altruistically and freely whether to vaccinate their children. We model eager and reluctant vaccinationist parents. We provide numerical evidence suggesting that individual anticipatory behavior may lead to a transient increase in measles prevalence before steady state eradication. This would cause non negligible welfare transfers between generations. Ironically, in our scenario, reluctant vaccinationists are among those who benefit the most from mandatory vaccination.

Personalized oncology with artificial intelligence: The case of temozolomide.

PURPOSE: Using artificial intelligence techniques, we compute optimal personalized protocols for temozolomide administration in a population of patients with variability. METHODS: Our optimizations are based on a Pharmacokinetics/Pharmacodynamics (PK/PD) model with population variability for temozolomide, inspired by Faivre et al. [10] and Panetta et al. [25,26]. The patient pharmacokinetic parameters can only be partially observed at admission and are progressively learned by Bayesian inference during treatment. For every patient, we seek to minimize tumor size while avoiding severe toxicity, i.e. maintaining an acceptable toxicity level. The optimization algorithm we rely on borrows from the field of artificial intelligence. RESULTS: Optimal personalized protocols (OPP) achieve a sizable decrease in tumor size at the population level but also patient-wise. The tumor size is on average 67.2 g lighter than with the standard maximum-tolerated dose protocol (MTD) after 336 days (12 MTD cycles). The corresponding 90% confidence interval for average tumor size reduction amounts to 58.6-82.7 g. When treated with OPP, less patients experience severe toxicity in comparison to MTD. MAJOR FINDINGS: We quantify in-silico the benefits offered by personalized oncology in the case of temozolomide administration. To do so, we compute optimal personalized protocols for a population of heterogeneous patients using artificial intelligence techniques. At each treatment day, the protocol is updated by taking into account the feedback obtained from patient's reaction to the drug administration. Personalized protocols greatly differ from each other, and from the standard MTD protocol. Benefits of personalization are very sizable: tumor sizes are much smaller on average and also patient-wise, while severe toxicity is made less frequent.

Optimizing treatment combination for lymphoma using an optimization heuristic.

BACKGROUND: The standard treatment for high-grade non-Hodgkin lymphoma involves the combination of chemotherapy and immunotherapy. We characterize in-silico the optimal combination protocol that maximizes the overall survival probability. We rely on a pharmacokinetics/pharmacodynamics (PK/PD) model that describes the joint evolution of tumor and effector cells, as well as the effects of both chemotherapy and immunotherapy. The toxicity is taken into account through ad-hoc constraints. We develop an optimization algorithm that belongs to the class of Monte-Carlo tree search algorithms. Our simulations rely on an in-silico population of heterogeneous patients differing with respect to their PK/PD parameters. The optimization objective consists in characterizing the combination protocol that maximizes the overall survival probability of the patient population under consideration. RESULTS: We compare using in-silico experiments our results to standard protocols and observe a gain in overall survival probabilities that vary from 4 to 9 percentage points. The gains increase with the complexity of the potential protocol. Gains are larger in presence of a higher number of injections or of an actual combination with immunotherapy. CONCLUSIONS: In in-silico experiments, optimal protocols achieve significant gains over standard protocols when considering overall survival probabilities. Our optimization algorithm enables us to efficiently tackle this numerical problem with a large dimensionality. The in-vivo implications of our in-silico results remain to be explored.

Optimizing immune cell therapies with artificial intelligence.

PURPOSE: We determine an optimal injection pattern for anti-vascular endothelial growth factor (VEGF) and for the combination of anti-VEGF and unlicensed dendritic cells. METHODS: We rely on the mathematical model of Soto-Ortiz and Finley (2016) for the interactions between the tumor growth, angiogenesis and immune system reactions. Our optimization algorithm belongs to the class of Monte-Carlo tree search algorithms. The objective consists in finding the minimal total drug doses for which an injection pattern yields tumor eradication. RESULTS: Our results are twofold. First, optimized injection protocols enable to significantly reduce the total drug dose for tumor elimination. For instance, for an early diagnosis date, a total dose equal to 58% of the standard anti-VEGF dose enables to eliminate the tumor. In the case of drug combination, associating 25% of the total standard anti-VEGF dose to 10% of the dendritic cell total standard dose eradicates tumor. Our second result is that administering a dose equal to the maximal standard dose allows for later diagnosis date compared to standard protocol. For instance, in the case of anti-VEGF injection, the optimal protocol postpones the maximal diagnosis date by more than one month. CONCLUSIONS: Overall, our optimization based on artificial intelligence delivers significant gains in total drug administration or in the length of the therapeutic window. Our method is flexible and could be adapted to other drug combinations.

Optimal dynamic regimens with artificial intelligence: The case of temozolomide.

We determine an optimal protocol for temozolomide using population variability and dynamic optimization techniques inspired by artificial intelligence. We use a Pharmacokinetics/Pharmacodynamics (PK/PD) model based on Faivre and coauthors (Faivre, et al., 2013) for the pharmacokinetics of temozolomide, as well as the pharmacodynamics of its efficacy. For toxicity, which is measured by the nadir of the normalized absolute neutrophil count, we formalize the myelosuppression effect of temozolomide with the physiological model of Panetta and coauthors (Panetta, et al., 2003). We apply the model to a population with variability as given in Panetta and coauthors (Panetta, et al., 2003). Our optimization algorithm is a variant in the class of Monte-Carlo tree search algorithms. We do not impose periodicity constraint on our solution. We set the objective of tumor size minimization while not allowing more severe toxicity levels than the standard Maximum Tolerated Dose (MTD) regimen. The protocol we propose achieves higher efficacy in the sense that -compared to the usual MTD regimen- it divides the tumor size by approximately 7.66 after 336 days -the 95% confidence interval being [7.36-7.97]. The toxicity is similar to MTD. Overall, our protocol, obtained with a very flexible method, gives significant results for the present case of temozolomide and calls for further research mixing operational research or artificial intelligence and clinical research in oncology.

Administration of temozolomide: Comparison of conventional and metronomic chemotherapy regimens.

PURPOSE: We compare the Maximum Tolerated Dose (MTD) and Metronomic Chemotherapy (MC) protocols for temozolomide administration. We develop an innovative methodology for characterizing optimal chemotherapy regimens. METHODS: We use a PK/PD model based on Faivre et al. (2013) for the pharmacokinetics of temozolomide, as well as the pharmacodynamics of its efficacy. For toxicity, which is measured by the nadir of the normalized absolute neutrophil count, we formalize the myelosuppression effect of temozolomide with the physiological model of Panetta et al. (2003b). We introduce a multi-criteria tool for comparing protocols along their efficacy and toxicity dimensions. RESULTS: We show that the toxicity of the MC regimen proposed by Faivre et al. (2013) can greatly be reduced without affecting its efficacy, while the standard MTD protocol efficacy cannot be improved without impairing its toxicity. We also show that for any acceptable toxicity level, the optimal protocol remains closely related to standard MTD. CONCLUSIONS: Overall, our new method enables a rich comparison between protocols along multiple dimensions. We can rank protocols for temozolomide administration. It is a first step toward building optimal individual protocols.

Canonical modeling of anticipatory vaccination behavior and long term epidemic recurrence.

Vaccination is one of humanity's main tools to fight epidemics. In most countries and for most diseases, vaccination is offered on a voluntary basis. Hence, the spread of a disease can be described as two interacting opposite dynamic systems: contagion is determined by past vaccination, while individuals decide whether to vaccinate based on beliefs regarding future disease prevalence. In this study, we show how the interplay between such anticipating behavior and the otherwise biological dynamics of a disease may lead to the emergence of recurrent patterns. We provide simulation results for (i) a Measles-like outbreak, (ii) canonical fully rational and far-sighted individuals, (iii) waning vaccine efficacy and vital dynamics, and (iv) long periods of time, i.e. long enough to observe several vaccination peaks. For comparison, we conducted a similar analysis for individuals with adaptive behavior. As an extension, we investigated the case where part of the population has an anti-vaccination stance.

The case for periodic OPV routine vaccination campaigns.

The possibility of periodic routine vaccination campaigns (PRVCs) is introduced in the context of a search for optimal oral poliovirus vaccine (OPV) administration strategies. Like the usual continuous routine vaccination campaign (CRVC), PRVCs target only newborns. However, they are not necessarily implemented continuously in time. Using a dynamic and compartmental polio transmission model in a stochastic context, it is shown that some PRVCs can achieve much greater efficiency than CRVC in terms of probability of wild poliovirus (WPV) eradication, even though they never outperform CRVC in terms of total number of paralytic infections. Moreover, these PRVCs results can be obtained at a lower price than CRVC. It is also shown that, even though PRVCs do not perform better than pulse vaccination campaigns (PVCs) when only epidemiological outputs are valued, they can do so when a cost-benefit analysis is preferred.

The probability of undetected wild poliovirus circulation: Can we do better?

Acute flaccid paralysis surveillance actively detects new paralytic infections caused by wild poliovirus (WPV). However, most WPV infections occur with no symptom. This complicates determining when WPV is eradicated in the context of stopping oral poliovirus vaccine (OPV). Previous studies have used the time since the last paralytic infection as a variable of interest to construct this probability. In this study, we show that more freely available information can be used. In particular, we focus on enriching the computation of the probability of WPV silent circulation with the date of occurrence of the last paralytic infection. We show that this information can for at least one set of conditions have crucial importance for an accurate estimation of the risk of false positive when declaring WPV eradicated. We also look at the importance of this information for optimal dynamic vaccination strategies.

Drug Launch Timing and International Reference Pricing.

This paper analyzes the timing decisions of pharmaceutical firms to launch a new drug in countries involved in international reference pricing. We show three important features of launch timing when all countries refer to the prices in all other countries and in all previous periods of time. First, there is no withdrawal of drugs in any country and in any period. Second, whenever the drug is sold in a country, it is also sold in all countries with larger willingness to pay. Third, there is no strict incentive to delay the launch of a drug in any country. We then show that the first and third results continue to hold when the countries only refer to the prices of a subset of all countries in a transitive way and in any period. We also show that the second result continues to hold when the reference is on the last period prices only. Last, we show that the seller's profits increase as the sets of reference countries decrease with respect to inclusion.

Evolution of cooperation with similarity to an archetype.

We use the framework of Colman with a Prisoner's Dilemma game and an evolutionary agent-based algorithm in order to study the evolution of cooperation and discrimination. We assume that players can discriminate on the basis of the phenotypic distance to an archetype, linked itself with a given behaviour in the game. However, we do not impose that the archetype corresponds to a conditionally cooperative behaviour. We show that cooperation can become the norm and discrimination can evolve spontaneously with no other assumption. For some archetypes, cooperation can even evolve faster and with more intensity than in the similarity-based case studied in Colman et al., 2012.