Using Explainable AI (XAI) for the Prediction of Falls in the Older Population

Abstract

The prevention of falls in older people requires the identification of the most important risk factors. Frailty is associated with risk of falls, but not all falls are of the same nature. In this work, we utilised data from The Irish Longitudinal Study on Ageing to implement Random Forests and Explainable Artificial Intelligence (XAI) techniques for the prediction of different types of falls and analysed their contributory factors using 46 input features that included those of a previously investigated frailty index. Data of participants aged 65 years and older were fed into four random forest models (all falls or syncope, simple fall, complex fall, and syncope). Feature importance rankings were based on mean decrease in impurity, and Shapley additive explanations values were calculated and visualised. Female sex and a previous fall were found to be of high importance in all of the models, and polypharmacy (being on five or more regular medications) was ranked high in the syncope model. The more ‘accidental’ (extrinsic) nature of simple falls was demonstrated in its model, where the presence of many frailty features had negative model contributions. Our results highlight that falls in older people are heterogenous and XAI can provide new insights to help their prevention.

Eight-year longitudinal falls trajectories and associations with modifiable risk factors: evidence from The Irish Longitudinal Study on Ageing (TILDA)

“…In addition, due to low numbers we did not study more granular transitions in falls subtypes (e.g. injurious, with recalled loss of consciousness) [ 27 ]. Our findings may not be generalisable and should be validated in clinical populations and external datasets.…”

“…The inclusion of TUG, MoCA and OH is due to previous work that described a ‘Bermuda triangle’ of falls risk in older people, characterised by gait disorder, cognitive impairment, and postural hypotension [ 25 , 26 ]. Other covariates were chosen based on previous research and an emphasis on including potentially modifiable factors [ 27–30 ]. All covariates were the values from Wave 1 assessment and remained constant across waves.…”

CNN-Based Self-Attention Weight Extraction for Fall Event Prediction Using Balance Test Score

“…Indeed, explaining the predictions made by AI algorithms is mandatory for healthcare professionals and patients in order to check medical hypotheses and conduct assessments based on AI-powered clinical decision support systems [12]. This is particularly important in fall event prediction, where healthcare professionals need to understand the factors that contribute to an individual's risk of falling in order to develop effective prevention strategies [13].…”

“…Yet, deep learning models show clear limitations when it comes to interpretability due to the complex calculus that they introduce and the depth of their layers [37]. As a result, explainable AI was introduced to address this issue [13]. Attention mechanism models, also known as self-attention (SA) models, utilize an architecture that creates similarity links inside an input sequence in order to assign importance weights to its components [38].…”