Causal inference is a widely recognized concept in various domains, including medicine, for estimating the effect of a medication on a certain disease. During this estimation, overlap is commonly used to eliminate the error caused by other features. However, finding the real overlap region in practice is challenging due to the limited sample size and unknown data distribution. Therefore, some machine-learning methods have been proposed to estimate the overlap region. One such method is Overrule, a Python package proposed by Oberst et al. Overrule is based on rule-based lassification and estimates the overlap region by interpreting it as several rules across the features. However, it is still unclear how Overrule performs under different circumstances. Thus, the primary bjective of this project is to test the performance of Overrule with different datasets. To accomplish this, a series of tests are built and executed to evaluate the performance of Overrule in diverse scenarios.

The purpose of this research is to analyze the performance of Propensity Score Matching, a causal inference method for causal effect estimation. More specifically, investigate how Propensity Score Matching reacts to breaking the unconfoundedness assumption, one of its core conceptual pillars. This has been achieved by running PSM on synthetic data that upholds the unconfoundedness condition, and then comparing these results with measurements obtained from running the algorithm on data with confounding features with varying contribution to other variable values and hiding these features individually or in progressively higher numbers. These results are also then compared to Linear Regression, a generic machine learning algorithm, for the sake of comparison of performance. The results obtained point to the observation that when hiding variables that only contribute to the main effect, treatment effect or treatment propensity calculation respectively, PSM performs with the same error no matter which of the three effects the hidden feature affects, making them equivalent in their error contribution. Additionally, it has also become apparent that in all experimental scenarios used in this work, PSM performed very similarly to Linear Regression and did not seem to offer any advantages over the latter in these specific situations.

Background and aims The timing of extubation is a difficult decision for the medical team on the PICU. With negative impact on patient outcome when extubating too late or too early. The aim of this study was to create machine learning models for extubation failure prediction after surgery in patients with congenital heart disease. The goal was to assess the influence of time variant features on the performance. Methods Data from post cardiac surgery patients admitted to the PICU of the University Medical Centre Utrecht, The Netherlands, between 2009 and 2018 was collected. Ventilator and monitor parameters were extracted in 12-hour segments. Different representations of time-variant features were calculated (per hour/ per 12-hour segment), these representations were tested against machine learning trained on only time-invariant features (age, weight diagnosis). Machine learning algorithms tested were: long short-term memory network (LSTM), logistic regression and random forest model. Models were evaluated by comparing the areas under the receiver operator curves Results With only time invariant features a performance of 75% [95%CI 81%-90%] using logistic regression. Adding the time-variant features to a LSTM model a performance was reached 77% [95%CI 80%-90%]. Important features from the logistic regression models were age, weight, heart rate and respiratory rate. Conclusions Based on the overall results we concluded that the chosen representations of time variant features did not significantly improve the performance of the models. To improve performance and implementation of machine learning models in the future, transparent and externally validated models need to be developed.