ABSTRACT
Urinary Tract Infections (UTIs) are one of the most prevalent bacterial infections in older adults and a significant contributor to unplanned hospital admissions in People Living with Dementia (PLWD), with early detection being crucial due to the predicament of reporting symptoms and limited help-seeking behaviour. The most common diagnostic tool is urine sample analysis, which can be time-consuming and is only employed where UTI clinical suspicion exists. In this method development and proof-of-concept study, participants living with dementia were monitored via low-cost devices in the home that passively measure activity, sleep, and nocturnal physiology. Using 27828 person-days of remote monitoring data (from 117 participants), we engineered features representing symptoms used for diagnosing a UTI. We then evaluate explainable machine learning techniques in passively calculating UTI risk and perform stratification on scores to support clinical translation and allow control over the balance between alert rate and sensitivity and specificity. The proposed UTI algorithm achieves a sensitivity of 65.3% (95% Confidence Interval (CI) = 64.3-66.2) and specificity of 70.9% (68.6-73.1) when predicting UTIs on unseen participants and after risk stratification, a sensitivity of 74.7% (67.9-81.5) and specificity of 87.9% (85.0-90.9). In addition, feature importance methods reveal that the largest contributions to the predictions were bathroom visit statistics, night-time respiratory rate, and the number of previous UTI events, aligning with the literature. Our machine learning method alerts clinicians of UTI risk in subjects, enabling earlier detection and enhanced screening when considering treatment.
ABSTRACT
BACKGROUND: People living with dementia (PLWD) have an increased susceptibility to developing adverse physical and psychological events. Internet of Things (IoT) technologies provides new ways to remotely monitor patients within the comfort of their homes, particularly important for the timely delivery of appropriate healthcare. Presented here is data collated as part of the on-going UK Dementia Research Institute's Care Research and Technology Centre cohort and Technology Integrated Health Management (TIHM) study. There are two main aims to this work: first, to investigate the effect of the COVID-19 quarantine on the performance of daily living activities of PLWD, on which there is currently little research; and second, to create a simple classification model capable of effectively predicting agitation risk in PLWD, allowing for the generation of alerts with actionable information by which to prevent such outcomes. METHOD: A within-subject, date-matched study was conducted on daily living activity data using the first COVID-19 quarantine as a natural experiment. Supervised machine learning approaches were then applied to combined physiological and environmental data to create two simple classification models: a single marker model trained using ambient temperature as a feature, and a multi-marker model using ambient temperature, body temperature, movement, and entropy as features. RESULT: There are 102 PLWD total included in the dataset, with all patients having an established diagnosis of dementia, but with ranging types and severity. The COVID-19 study was carried out on a sub-group of 21 patient households. In 2020, PLWD had a significant increase in daily household activity (p = 1.40e-08), one-way repeated measures ANOVA). Moreover, there was a significant interaction between the pandemic quarantine and patient gender on night-time bed-occupancy duration (p = 3.00e-02, two-way mixed-effect ANOVA). On evaluating the models using 10-fold cross validation, both the single and multi-marker model were shown to balance precision and recall well, having F1-scores of 0.80 and 0.66, respectively. CONCLUSION: Remote monitoring technologies provide a continuous and reliable way of monitoring patient day-to-day wellbeing. The application of statistical analyses and machine learning algorithms to combined physiological and environmental data has huge potential to positively impact the delivery of healthcare for PLWD.
