Treffer: Advanced statistics analysis on Finnish Maternity Cohort and Swedish Cervical Screening Cohort (part A). Public Deliverable 6.5

The research question is whether we can use detailed cervical screeningdata to predict the risk for cervical cancer and the risk for high-gradelesions. The broad analytical approach is an adaptation of the screeningmodels developed by Day and Walter during the 1980s. To our knowledge,our approach is novel for cervical cancer screening. We provide (a) SQL andSAS code for the data extraction from the Swedish Cervical CancerScreening Register and (b) R and C++ code for the likelihood construction,optimisation, and predictions. As a proof of concept, we fit the model to thecohort of women born in 1960 who were living in Sweden on their fifteenthbirthday. We found some issues with fitting the model due to the lack ofidentifiability of the model parameters. We propose some extensions to thecurrent approach, including scaling up the computations to more birthcohorts and extending the model to include negative biopsies. Finally, weconclude that such a mathematical approach could be compared withpredictions based on machine learning and evaluate whether the machinelearning algorithm gives sufficient weight to different screening histories.