Search

Data Study Groups are week-long events at The Alan Turing Institute bringing together some of the country’s top talent from data science, artificial intelligence, and wider fields, to analyse real-world data science challenges.\ud \ud Roche: Personalised lung cancer treatment modelling using electronic health records and genomics\ud \ud Cancer immunotherapy (CIT) is a promising new type of cancer treatment that uses the patient’s own immune system to fight cancer cells. CIT drugs work to stop the cancer cells from turning off the immune system’s T-cells by inhibiting the PD-L1 produced by the tumour cells (PD-L1 is a protein that binds to PD-1 receptors on T-cells and prevents the immune system from attacking the cancer cells).\ud \ud CIT is currently being used to treat patients with non-small cell lung cancer (NSCLC) for whom chemotherapy or other drugs have failed. CIT is also be-ing used as part of the first-line treatment in patients with advanced NSCLC (aNSCLC - stage III and higher). Theoretically, patients with high PD-L1 ex-pression levels are more likely to respond well to CIT; however, in practice, patient outcomes vary considerably.\ud \ud In this data study group, we investigated different approaches for predicting survival time for patients treated with CIT as first line of treatment, using both electronic health records and tumour genomic data. We also investigated the causal effects of CIT vs other oncology treatments, and studied treatment heterogeneity. The results contribute to identifying patients who are most likely to benefit from CIT.

The Finite-Difference Time-Domain (FDTD) method is a numerical analysis technique used for the simulation of electromagnetic (EM) phenomena. It is one of the most popular computational techniques for modelling the behaviour of electromagnetic waves. As an explicit method, the FDTD scheme is constrained by the Courant-Friedrichs-Lewy (CFL) stability condition, which limits the time step that can be used to obtain stable and accurate simulation results. It can be an important disadvantage for modelling specific scenarios such as large computational EM problems or problems where very fine meshes, compared to the electrical wavelength required. This leads to longer simulation time requiring larger computational requirements. In order to overcome this problem, implicit FDTD methods are an alternative. Numerous researches have been carried out so as to develop schemes to provide the computational stability beyond the limit of the CFL condition. In this Thesis, an Additive Operator Splitting (AOS) FDTD method has been introduced.

We explore increases in the nominal minimum wage in a difference-in-differences setting to estimate match survival wage elasticity. The elasticity is negative and larger than one for matches directly affected by minimum wage increases, those with paying below the new minimum wage. The impact of the minimum wage is stronger for young workers and the manufacturing sector, groups with a large and increasing share of low-wage workers. Given the low-wage mobility in the Portuguese labor market and the large share of workers directly affected by the minimum wage increase (reaching almost 18%), these results call for a careful analysis of the set of policies that interfere with low-wage employment.

In a segmented labor market, theory predicts that employment protection has an asymmetric impact on entry and incumbent wages. We explore a reform that increased the protection of open-ended contracts for a well-defined subset of firms, while leaving it unchanged for other firms. The causal evidence points to a reduction in wages for new open-ended and fixed-term contracts and no impact for more tenured workers. The reductions estimated for entrants oscillate between -0.9 and -0.5 p.p., covering a significant part of the expected increase in firing costs. Firms with larger shares of fixed-term contracts shifted the burden to these workers.

The fact that unemployed workers have different abilities to smooth consumption entails heterogeneous responses to extended unemployment benefits. Our empirical exercise explores a quasi-experimental setting generated by an increase in the benefits entitlement period. The results point towards a hump-shape response of unemployment duration over the one-year pre-unemployment wage distribution; individuals at the bottom and at the top of the wage distribution reacted less than those in the interquartile range. This behavior of job searchers is consistent with labor supply models with unemployment insurance and savings. It questions the optimality of very long entitlement periods to target the unemployment experiences of low-wage workers.

Portuguese firms engage in intense reallocation, most employers simultaneously hire and separate from workers, resulting in a large heterogeneity of flows and excess turnover. Large and older firms have lower flows, but high excess turnover rates. In small firms, hires and separations move symmetrically during expansion and contraction periods, on the contrary, large firms adjust their employment levels by reducing entry and not by increasing separations. Most hires and separations are on fixed-term contracts and shrinking firms replace a larger share of their separations under fixed-term contracts, while expanding firms replace most of the separations under open-ended contracts. The comparison with the U.S. shows that while worker and job flows are lower in Portugal, the excess turnover is remarkably close in the two countries.