Focus on an infrequently used quantity in the context of competing risks: The conditional probability function

The role of depression in the association between mobilisation timing and live discharge after hip fracture surgery: Secondary analysis of the UK National Hip Fracture Database

PURPOSE

The aim was to compare the probability of discharge after hip fracture surgery conditional on being alive and in hospital between patients mobilised within and beyond 36-hours of surgery across groups defined by depression.

METHODS

Data were taken from the National Hip Fracture Database and included patients 60 years of age or older who underwent hip fracture surgery in England and Wales between 2014 and 2016. The conditional probability of postsurgical live discharge was estimated for patients mobilised early and for patients mobilised late across groups with and without depression. The association between mobilisation timing and the conditional probability of live discharge were also estimated separately through adjusted generalized linear models.

RESULTS

Data were analysed for 116,274 patients. A diagnosis of depression was present in 8.31% patients. In those with depression, 7,412 (76.7%) patients mobilised early. In those without depression, 84,085 (78.9%) patients mobilised early. By day 30 after surgery, the adjusted odds ratio of discharge among those who mobilised early compared to late was 1.79 (95% CI: 1.56-2.05, p<0.001) and 1.92 (95% CI: 1.84-2.00, p<0.001) for those with and without depression, respectively.

CONCLUSION

A similar proportion of patients with depression mobilised early after hip fracture surgery when compared to those without a diagnosis of depression. The association between mobilisation timing and time to live discharge was observed for patients with and without depression.

MS-CPFI: A model-agnostic Counterfactual Perturbation Feature Importance algorithm for interpreting black-box Multi-State models

Multi-state processes (Webster, 2019) are commonly used to model the complex clinical evolution of diseases where patients progress through different states. In recent years, machine learning and deep learning algorithms have been proposed to improve the accuracy of these models' predictions (Wang et al., 2019). However, acceptability by patients and clinicians, as well as for regulatory compliance, require interpretability of these algorithms's predictions. Existing methods, such as the Permutation Feature Importance algorithm, have been adapted for interpreting predictions in black-box models for 2-state processes (corresponding to survival analysis). For generalizing these methods to multi-state models, we introduce a novel model-agnostic interpretability algorithm called Multi-State Counterfactual Perturbation Feature Importance (MS-CPFI) that computes feature importance scores for each transition of a general multi-state model, including survival, competing-risks, and illness-death models. MS-CPFI uses a new counterfactual perturbation method that allows interpreting feature effects while capturing the non-linear effects and potentially capturing time-dependent effects. Experimental results on simulations show that MS-CPFI increases model interpretability in the case of non-linear effects. Additionally, results on a real-world dataset for patients with breast cancer confirm that MS-CPFI can detect clinically important features and provide information on the disease progression by displaying features that are protective factors versus features that are risk factors for each stage of the disease. Overall, MS-CPFI is a promising model-agnostic interpretability algorithm for multi-state models, which can improve the interpretability of machine learning and deep learning algorithms in healthcare.