Osteosarcoma Survivorship

Relationship between individual received dose intensity and survival

Received dose intensity is a keystone concept in chemotherapy. Roughly speaking is the amount of drugs delivered or planned in a certain time window. Traditionally it is computed on the chemotherapy schedule before treatment. This target value of received dose intensity is therefore the same for patients who share the allocated regimen. In contrast, achieved received dose intensity is different for each patient, because it is calculated on the actual dose received and the actual treatment duration. Finally, regulated received dose intensity extends the achieved calculation to a longitudinal setting by embedding patients' individual tolerability.

This project aims at answering the following questions:

• To what extent does target received dose intensity predict the achieved one?
• Is there any difference in the associations between target, achieved, or regulated received dose intensity and survival?
• Is received dose intensity a good risk factor for survival in osteosarcoma?
• What is the role of individual tolerability to chemotherapy in survivorship?

Survival Analysis in presence of treatment-adjustment bias

The effect of chemotherapy on survival outcome is difficult to estimate in an unbiased way. Chemotherapy-induced toxicities are risk factors for survival and predictors of treatment adaptations (delays and/or dose reductions) at the same time. In other words, toxicities are time-dependent confounders. Inverse Probability of Treatment Weighting is a statistical technique that rebalance treatment adaptations in each strata of the confounders at all time points. In this way, treatment adaptations can be made independent of toxicities and a randomised controlled trial can be mimicked. This technique allows for unbiased estimation of the causal effect of treatment adaptations on survival outcome.

References

Joshua Feinberg, Claudia Ashton, Allison Hirst, Christopher Pennell, Peter McCulloch. Meeting abstracts from the 4th International Clinical Trials Methodology Conference (ICTMC) and the 38th Annual Meeting of the Society for Clinical Trials . Trials. 2017; 18(suppl 1):200.

Anninga J, Lancia C, Spitoni C, Whelan J, Sydes MR, Jovic G, Gelderblom H, Hogendoorn PCW, Fiocco M. Causal effects of methotrexate reductions/delays in treatment of high-grade resectable osteosarcoma. Journal of Clinical Oncology, 34(suppl 15).

Pre-Scheduled Random Arrivals and Air-Traffic Applications

Pre-Scheduled Random Arrivals are a class of point processes obtained by superimposing random independent fluctuations to a fixed schedule of customers. This model is particularly suited for the transportation filed, where arrivals are intrinsically subject to random delays. Derivation of analytical results is possible yet very tough. Applied studies show that the model is very fit for describing the actual inbound flow of aircraft at a large airport.

References

Lancia C, Lulli G. Asymptotics for the Late Arrivals Problem. 2017. ArXiV Preprint arxiv.org/abs/1708.02486

Lancia C, Guadagni G, Ndreca S, and Scoppola B. Asymptotics for the Late Arrivals Problem. 2017. ArXiV Preprint arxiv.org/abs/1302.1999

Caccavale MV, Iovanella A, Lancia C, Lulli G, Scoppola B. A model of inbound air traffic: The application to Heathrow airport. Journal of Air Transport Management. 2014; 34: 116-122.

Probabilistic Cellular Automata

Probabilistic celllular automata are Markov chains with a parallel update rule. This means that the states of the chain can be decomposed in single agents and that the chain transition probabilities can be factorized over the agents. These models are of key importance in the study of non-equilibrium stationary states. They are also interesting for application, for example in vehicular traffic.

References

Lancia C, Scoppola B. Ising Model on the Torus and PCA Dynamics: reversibility, irreversibility, and fast tunneling. In Louis, P.Y. and Nardi, F.R. (editors), Probabilistic Cellular Automata: Theory, Applications and Future Perspectives. Springer: 2018; Chapter 7.

Scoppola B, Lancia C, Mariani R. On the Blockage Problem and the Non-analyticity of the Current for Parallel TASEP on a Ring. Journal of Statistical Physics. 2015; 161(4): 843-858.

Lancia C, Scoppola B. Equilibrium and non-equilibrium Ising models by means of PCA. Journal of Statistical Physics. 2013; 153(4): 641-653.

ProGA

ProGA was a SESARJU WP-E project that developed the the idea of estimating the near-future location of general aviation traffic. The project was motivated by a substantial gap in safety performance in commercial vs general aviation, with estimated accident rates of 6 and 100 accidents per 10 million flight hours, respectively. ProGA was developed by a consortium including Deep Blue s.r.l. (Rome, IT), Onera (Paris, FR), and NLR (Amsterdam, NL).

References

Lancia, C., Taurino, D., Frau, G., Verstraeten, J., Le Tallec, C., 2014. Traffic Predictions Supporting General Aviation. In Schaefer, D. (Editor) Proceedings of the SESAR Innovation Days, EUROCONTROL.

Looking Through the Cutoff Window

A family of Markov chains exhibits cutoff whenever it presents a sharp, abrupt transition in the distance from equilibrium. Study of this phenomenon is important to characterize the mixing time of the chain. The contribution of the project is to highlight the contributions to the time needed to complete the transition towards the equilibrium state (the so-called cutoff window).

References

Lancia C, Nardi FR, Scoppola B. Entropy-driven cutoff phenomena. Journal of Statistical Physics. 2012; 149(1): 108-141.