RESUMO
OBJECTIVE: To test the ability of high-performance machine learning (ML) models employing clinical, radiological, and radiomic variables to improve non-invasive prediction of the pathological status of prostate cancer (PCa) in a large, single-institution cohort. METHODS: Patients who underwent multiparametric MRI and prostatectomy in our institution in 2015-2018 were considered; a total of 949 patients were included. Gradient-boosted decision tree models were separately trained using clinical features alone and in combination with radiological reporting and/or prostate radiomic features to predict pathological T, pathological N, ISUP score, and their change from preclinical assessment. Model behavior was analyzed in terms of performance, feature importance, Shapley additive explanation (SHAP) values, and mean absolute error (MAE). The best model was compared against a naïve model mimicking clinical workflow. RESULTS: The model including all variables was the best performing (AUC values ranging from 0.73 to 0.96 for the six endpoints). Radiomic features brought a small yet measurable boost in performance, with the SHAP values indicating that their contribution can be critical to successful prediction of endpoints for individual patients. MAEs were lower for low-risk patients, suggesting that the models find them easier to classify. The best model outperformed (p ≤ 0.0001) clinical baseline, resulting in significantly fewer false negative predictions and overall was less prone to under-staging. CONCLUSIONS: Our results highlight the potential benefit of integrative ML models for pathological status prediction in PCa. Additional studies regarding clinical integration of such models can provide valuable information for personalizing therapy offering a tool to improve non-invasive prediction of pathological status. CLINICAL RELEVANCE STATEMENT: The best machine learning model was less prone to under-staging of the disease. The improved accuracy of our pathological prediction models could constitute an asset to the clinical workflow by providing clinicians with accurate pathological predictions prior to treatment. KEY POINTS: ⢠Currently, the most common strategies for pre-surgical stratification of prostate cancer (PCa) patients have shown to have suboptimal performances. ⢠The addition of radiological features to the clinical features gave a considerable boost in model performance. Our best model outperforms the naïve model, avoiding under-staging and resulting in a critical advantage in the clinic. â¢Machine learning models incorporating clinical, radiological, and radiomics features significantly improved accuracy of pathological prediction in prostate cancer, possibly constituting an asset to the clinical workflow.
RESUMO
We analyze the role of disease containment policy in the form of treatment in a stochastic economic-epidemiological framework in which the probability of the occurrence of random shocks is state-dependent, namely it is related to the level of disease prevalence. Random shocks are associated with the diffusion of a new strain of the disease which affects both the number of infectives and the growth rate of infection, and the probability of such shocks realization may be either increasing or decreasing in the number of infectives. We determine the optimal policy and the steady state of such a stochastic framework, which is characterized by an invariant measure supported on strictly positive prevalence levels, suggesting that complete eradication is never a possible long run outcome where instead endemicity will prevail. Our results show that: (i) independently of the features of the state-dependent probabilities, treatment allows to shift leftward the support of the invariant measure; and (ii) the features of the state-dependent probabilities affect the shape and spread of the distribution of disease prevalence over its support, allowing for a steady state outcome characterized by a distribution alternatively highly concentrated over low prevalence levels or more spread out over a larger range of prevalence (possibly higher) levels.
RESUMO
The COVID-19 pandemic has illustrated the unprecedented challenges of ensuring the continuity of operations in a supply chain as suppliers' and their suppliers stop producing due the spread of infection, leading to a degradation of downstream customer service levels in a ripple effect. In this paper, we contextualize a dynamic approach and propose an optimal control model for supply chain reconfiguration and ripple effect analysis integrated with an epidemic dynamics model. We provide supply chain managers with the optimal choice over a planning horizon among subsets of interchangeable suppliers and corresponding orders; this will maximize demand satisfaction given their prices, lead times, exposure to infection, and upstream suppliers' risk exposure. Numerical illustrations show that our prescriptive forward-looking model can help reconfigure a supply chain and mitigate the ripple effect due to reduced production because of suppliers' infected workers. A risk aversion factor incorporates a measure of supplier risk exposure at the upstream echelons. We examine three scenarios: (a) infection limits the capacity of suppliers, (b) the pandemic recedes but not at the same pace for all suppliers, and (c) infection waves affect the capacity of some suppliers, while others are in a recovery phase. We illustrate through a case study how our model can be immediately deployed in manufacturing or retail supply chains since the data are readily accessible from suppliers and health authorities. This work opens new avenues for prescriptive models in operations management and the study of viable supply chains by combining optimal control and epidemiological models.
RESUMO
The COVID-19 pandemic wreaks havoc in supply chains by reducing the production capacity of some essential suppliers, closure of production facilities or the absence of infected workers. In this paper, we present three decision support models for a plant manager to help in deciding on (a) the level of protection of the workforce against the spread of the virus in the absence of regional protection measures, (b) on the duration of the protection, and (c) the level of protection of the workforce with regional protection measures enforced by health authorities. These decision models are based on a SIS epidemiological model which takes into account the possibility that a worker can infect others but also that even when recovered can be infected again. The first and third models prescribe how, in time, the protection effort in terms of prophylactic measures must be deployed. The second model extends the first one as it also determines the length the protection effort must be deployed. The proposed models have been applied to the case of a meat processing plant that must satisfy the demand of a large-scale retailer. Clearly, to achieve production targets and satisfy customers' demand, plants in this labor-intensive industry rely on the number of healthy workers and the service level of suppliers. Our results indicate that these models provide managers with the tools to understand and measure the impact of an infection on production and the corresponding cost. Along the way, this work illustrates the ripple effect as suppliers affected by the pandemic are unable to fulfill the processing plant requirements and so the retailer's orders. Our findings provide normative guidance for supply chain decision support systems under risk of pandemic induced disruptions using a quantitative model-based approach.
RESUMO
A probabilistic approach to the epidemic evolution on realistic social-contact networks allows for characteristic differences among subjects, including the individual number and structure of social contacts, and the heterogeneity of the infection and recovery rates according to age or medical preconditions. Within our probabilistic Susceptible-Infectious-Removed (SIR) model on social-contact networks, we evaluate the infection load or activation margin of various control scenarios; by confinement, by vaccination, and by their combination. We compare the epidemic burden for subpopulations that apply competing or cooperative control strategies. The simulation experiments are conducted on randomized social-contact graphs that are designed to exhibit realistic person-person contact characteristics and which follow near homogeneous or block-localized subpopulation spreading. The scalarization method is used for the multi-objective optimization problem in which both the infection load is minimized and the extent to which each subpopulation's control strategy preference ranking is adhered to is maximized. We obtain the compounded payoff matrices for two subpopulations that impose contrasting control strategies, each according to their proper ranked control strategy preferences. The Nash equilibria, according to each subpopulation's compounded objective, and according to their proper ranking intensity, are discussed. Finally, the interaction effects of the control strategies are discussed and related to the type of spreading of the two subpopulations.
RESUMO
We analyze the implications of strategic interactions between two heterogeneous groups (i.e., young and old, men and women) in a macroeconomic-epidemiological framework. The interactions between groups determine the overall prevalence of a communicable disease, which in turn affects the level of economic activity. Individuals may lower their disease exposure by reducing their mobility, but since changing mobility patterns is costly, each group has an incentive to free ride negatively affecting the chances of disease containment at the aggregate level. By focusing on an early epidemic setting, we explicitly characterize the cooperative and noncooperative equilibria, determining how the inefficiency induced by noncooperation (i.e., failure to internalize epidemic externalities) depends both on economic and epidemiological parameters. We show that long-run eradication may be possible even in the absence of coordination, but coordination leads to a faster reduction in the number of infectives in finite time. Moreover, the inefficiency induced by noncooperation increases (decreases) with the factors increasing (decreasing) the pace of the disease spread.
RESUMO
We analyze the determination of the optimal intensity and duration of social distancing policy aiming to control the spread of an infectious disease in a simple macroeconomic-epidemiological model. In our setting the social planner wishes to minimize the social costs associated with the levels of disease prevalence and output lost due to social distancing, both during and at the end of epidemic management program. Indeed, by limiting individuals' ability to freely move or interact with others (since requiring to wear face mask or to maintain physical distance from others, or even forcing some businesses to remain closed), social distancing has on the one hand the effect to reduce the disease incidence and on the other hand to reduce the economy's productive capacity. We analyze both the early and the advanced epidemic stage intervention strategies highlighting their implications for short and long run health and macroeconomic outcomes. We show that both the intensity and the duration of the optimal social distancing policy may largely vary according to the epidemiological characteristics of specific diseases, and that the balancing of the health benefits and economic costs associated with social distancing may require to accept the disease to reach an endemic state. Focusing in particular on COVID-19 we present a calibration based on Italian data showing how the optimal social distancing policy may vary if implemented at national or at regional level.
RESUMO
We analyze a discrete time two-sector economic growth model where the production technologies in the final and human capital sectors are affected by random shocks both directly (via productivity and factor shares) and indirectly (via a pollution externality). We determine the optimal dynamics in the decentralized economy and show how these dynamics can be described in terms of a two-dimensional affine iterated function system with probability. This allows us to identify a suitable parameter configuration capable of generating exactly the classical Barnsley's fern as the attractor of the log-linearized optimal dynamical system.