Search

Aerial image data are becoming more widely available, and analysis techniques based on supervised learning are advancing their use in a wide variety of remote sensing contexts. However, supervised learning requires training datasets which are not always available or easy to construct with aerial imagery. In this respect, unsupervised machine learning techniques present important advantages. This work presents a novel pipeline to demonstrate how available aerial imagery can be used to better the provision of services related to the built environment, using the case study of road traffic collisions (RTCs) across three cities in the UK. In this paper, we show how aerial imagery can be leveraged to extract latent features of the built environment from the purely visual representation of top-down images. With these latent image features in hand to represent the urban structure, this work then demonstrates how hazardous road segments can be clustered to provide a data-augmented aid for road safety experts to enhance their nuanced understanding of how and where different types of RTCs occur. [ABSTRACT FROM AUTHOR]

For efficient operation, solar power operators often require generation forecasts for multiple sites with varying data availability. Many proposed methods for forecasting solar irradiance / solar power production formulate the problem as a time-series, using current observations to generate forecasts. This necessitates a real-time data stream and enough historical observations at every location for these methods to be deployed. In this paper, we propose the use of Global methods to train generalised models. Using data from 20 locations distributed throughout the UK, we show that it is possible to learn models without access to data for all locations, enabling them to generate forecasts for unseen locations. We show a single Global model trained on multiple locations can produce more consistent and accurate results across locations. Furthermore, by leveraging weather observations and measurements from other locations we show it is possible to create models capable of accurately forecasting irradiance at locations without any real-time data. We apply our approaches to both classical and state-of-the-art Machine Learning methods, including a Transformer architecture. We compare models using satellite imagery or point observations (temperature, pressure, etc.) as weather data. These methods could facilitate planning and optimisation for both newly deployed solar farms and domestic installations from the moment they come online. [ABSTRACT FROM AUTHOR]

Much of the knowledge and information needed for enabling high-quality clinical research is stored in free-text format. Natural language processing (NLP) has been used to extract information from these sources at scale for several decades. This paper aims to present a comprehensive review of clinical NLP for the past 15 years in the UK to identify the community, depict its evolution, analyse methodologies and applications, and identify the main barriers. We collect a dataset of clinical NLP projects (n = 94; £ = 41.97 m) funded by UK funders or the European Union's funding programmes. Additionally, we extract details on 9 funders, 137 organisations, 139 persons and 431 research papers. Networks are created from timestamped data interlinking all entities, and network analysis is subsequently applied to generate insights. 431 publications are identified as part of a literature review, of which 107 are eligible for final analysis. Results show, not surprisingly, clinical NLP in the UK has increased substantially in the last 15 years: the total budget in the period of 2019–2022 was 80 times that of 2007–2010. However, the effort is required to deepen areas such as disease (sub-)phenotyping and broaden application domains. There is also a need to improve links between academia and industry and enable deployments in real-world settings for the realisation of clinical NLP's great potential in care delivery. The major barriers include research and development access to hospital data, lack of capable computational resources in the right places, the scarcity of labelled data and barriers to sharing of pretrained models. [ABSTRACT FROM AUTHOR]

The National COVID-19 Chest Imaging Database (NCCID) is a centralized UK database of thoracic imaging and corresponding clinical data. It is made available by the National Health Service Artificial Intelligence (NHS AI) Lab to support the development of machine learning tools focused on Coronavirus Disease 2019 (COVID-19). A bespoke cleaning pipeline for NCCID, developed by the NHSx, was introduced in 2021. We present an extension to the original cleaning pipeline for the clinical data of the database. It has been adjusted to correct additional systematic inconsistencies in the raw data such as patient sex, oxygen levels and date values. The most important changes will be discussed in this paper, whilst the code and further explanations are made publicly available on GitLab. The suggested cleaning will allow global users to work with more consistent data for the development of machine learning tools without being an expert. In addition, it highlights some of the challenges when working with clinical multi-center data and includes recommendations for similar future initiatives. [ABSTRACT FROM AUTHOR]

In this paper we characterize the performance of linear models trained via widely-used sparse machine learning algorithms. We build polygenic scores and examine performance as a function of training set size, genetic ancestral background, and training method. We show that predictor performance is most strongly dependent on size of training data, with smaller gains from algorithmic improvements. We find that LASSO generally performs as well as the best methods, judged by a variety of metrics. We also investigate performance characteristics of predictors trained on one genetic ancestry group when applied to another. Using LASSO, we develop a novel method for projecting AUC and correlation as a function of data size (i.e., for new biobanks) and characterize the asymptotic limit of performance. Additionally, for LASSO (compressed sensing) we show that performance metrics and predictor sparsity are in agreement with theoretical predictions from the Donoho-Tanner phase transition. Specifically, a future predictor trained in the Taiwan Precision Medicine Initiative for asthma can achieve an AUC of 0. 63 (0.02) and for height a correlation of 0. 648 (0.009) for a Taiwanese population. This is above the measured values of 0. 61 (0.01) and 0. 631 (0.008) , respectively, for UK Biobank trained predictors applied to a European population. [ABSTRACT FROM AUTHOR]

This paper presents a comprehensive dataset intended to evaluate passive Human Activity Recognition (HAR) and localization techniques with measurements obtained from synchronized Radio-Frequency (RF) devices and vision-based sensors. The dataset consists of RF data including Channel State Information (CSI) extracted from a WiFi Network Interface Card (NIC), Passive WiFi Radar (PWR) built upon a Software Defined Radio (SDR) platform, and Ultra-Wideband (UWB) signals acquired via commercial off-the-shelf hardware. It also consists of vision/Infra-red based data acquired from Kinect sensors. Approximately 8 hours of annotated measurements are provided, which are collected across two rooms from 6 participants performing 6 daily activities. This dataset can be exploited to advance WiFi and vision-based HAR, for example, using pattern recognition, skeletal representation, deep learning algorithms or other novel approaches to accurately recognize human activities. Furthermore, it can potentially be used to passively track a human in an indoor environment. Such datasets are key tools required for the development of new algorithms and methods in the context of smart homes, elderly care, and surveillance applications. Measurement(s) Human physical activity • Human location Technology Type(s) WiFi sensing device • ultra-wideband impulse radar • passive WiFi radar • Kinect motion sensor Factor Type(s) Human location • Human physical activity • Room geometry • Participant demographics • Contactless sensing devices Sample Characteristic - Organism Homo sapiens Sample Characteristic - Environment Office building Sample Characteristic - Location United Kingdom [ABSTRACT FROM AUTHOR]

Transferring and replicating predictive algorithms across healthcare systems constitutes a unique yet crucial challenge that needs to be addressed to enable the widespread adoption of machine learning in healthcare. In this study, we explored the impact of important differences across healthcare systems and the associated Electronic Health Records (EHRs) on machine-learning algorithms to predict mental health crises, up to 28 days in advance. We evaluated both the transferability and replicability of such machine learning models, and for this purpose, we trained six models using features and methods developed on EHR data from the Birmingham and Solihull Mental Health NHS Foundation Trust in the UK. These machine learning models were then used to predict the mental health crises of 2907 patients seen at the Rush University System for Health in the US between 2018 and 2020. The best one was trained on a combination of US-specific structured features and frequency features from anonymized patient notes and achieved an AUROC of 0.837. A model with comparable performance, originally trained using UK structured data, was transferred and then tuned using US data, achieving an AUROC of 0.826. Our findings establish the feasibility of transferring and replicating machine learning models to predict mental health crises across diverse hospital systems. [ABSTRACT FROM AUTHOR]

By September 2022, more than 600 million cases of SARS-CoV-2 infection have been reported globally, resulting in over 6.5 million deaths. COVID-19 mortality risk estimators are often, however, developed with small unrepresentative samples and with methodological limitations. It is highly important to develop predictive tools for pulmonary embolism (PE) in COVID-19 patients as one of the most severe preventable complications of COVID-19. Early recognition can help provide life-saving targeted anti-coagulation therapy right at admission. Using a dataset of more than 800,000 COVID-19 patients from an international cohort, we propose a cost-sensitive gradient-boosted machine learning model that predicts occurrence of PE and death at admission. Logistic regression, Cox proportional hazards models, and Shapley values were used to identify key predictors for PE and death. Our prediction model had a test AUROC of 75.9% and 74.2%, and sensitivities of 67.5% and 72.7% for PE and all-cause mortality respectively on a highly diverse and held-out test set. The PE prediction model was also evaluated on patients in UK and Spain separately with test results of 74.5% AUROC, 63.5% sensitivity and 78.9% AUROC, 95.7% sensitivity. Age, sex, region of admission, comorbidities (chronic cardiac and pulmonary disease, dementia, diabetes, hypertension, cancer, obesity, smoking), and symptoms (any, confusion, chest pain, fatigue, headache, fever, muscle or joint pain, shortness of breath) were the most important clinical predictors at admission. Age, overall presence of symptoms, shortness of breath, and hypertension were found to be key predictors for PE using our extreme gradient boosted model. This analysis based on the, until now, largest global dataset for this set of problems can inform hospital prioritisation policy and guide long term clinical research and decision-making for COVID-19 patients globally. Our machine learning model developed from an international cohort can serve to better regulate hospital risk prioritisation of at-risk patients. [ABSTRACT FROM AUTHOR]

In this study, we aimed to compare imaging-based features of brain function, measured by resting-state fMRI (rsfMRI), with individual characteristics such as age, gender, and total intracranial volume to predict behavioral measures. We developed a machine learning framework based on rsfMRI features in a dataset of 20,000 healthy individuals from the UK Biobank, focusing on temporal complexity and functional connectivity measures. Our analysis across four behavioral phenotypes revealed that both temporal complexity and functional connectivity measures provide comparable predictive performance. However, individual characteristics consistently outperformed rsfMRI features in predictive accuracy, particularly in analyses involving smaller sample sizes. Integrating rsfMRI features with demographic data sometimes enhanced predictive outcomes. The efficacy of different predictive modeling techniques and the choice of brain parcellation atlas were also examined, showing no significant influence on the results. To summarize, while individual characteristics are superior to rsfMRI in predicting behavioral phenotypes, rsfMRI still conveys additional predictive value in the context of machine learning, such as investigating the role of specific brain regions in behavioral phenotypes. Individual characteristics tend to over-perform fMRI-based metrics in the prediction of behavioral measures. However, the combination of both metrics can increase predictive accuracy, suggesting that fMRI-based metrics can successfully complement individual characteristics, mainly in large datasets. [ABSTRACT FROM AUTHOR]

The identification of metabolic biomarkers for aging-related diseases and mortality is of significant interest in the field of longevity. In this study, we investigated the associations between nuclear magnetic resonance (NMR) metabolomics biomarkers and aging-related diseases as well as mortality using the UK Biobank dataset. We analyzed NMR samples from approximately 110,000 participants and used multi-head machine learning classification models to predict the incidence of aging-related diseases. Cox regression models were then applied to assess the relevance of NMR biomarkers to the risk of death due to aging-related diseases. Additionally, we conducted survival analyses to evaluate the potential improvements of NMR in predicting survival and identify the biomarkers most strongly associated with negative health outcomes by dividing participants into health, disease, and death groups for all age groups. Our analysis revealed specific metabolomics profiles that were associated with the incidence of age-related diseases, and the most significant biomarker was intermediate density lipoprotein cholesteryl (IDL-CE). In addition, NMR biomarkers could provide additional contributions to relevant mortality risk prediction when combined with conventional risk factors, by improving the C-index from 0.813 to 0.833, with 17 NMR biomarkers significantly contributing to disease-related death, such as monounsaturated fatty acids (MUFA), linoleic acid (LA), glycoprotein acetyls (GlycA), and omega-3. Moreover, the value of free cholesterol in very large HDL particles (XL-HDL-FC) in the healthy control group demonstrated significantly higher values than the disease and death group across all age groups. This study highlights the potential of NMR metabolomics profiling as a valuable tool for identifying metabolic biomarkers associated with aging-related diseases and mortality risk, which could have practical implications for aging-related disease risk and mortality prediction. [ABSTRACT FROM AUTHOR]

Osteoarthritis (OA) is increasing in prevalence and has a severe impact on patients' lives. However, our understanding of biomarkers driving OA risk remains limited. We developed a model predicting the five-year risk of OA diagnosis, integrating retrospective clinical, lifestyle and biomarker data from the UK Biobank (19,120 patients with OA, ROC-AUC: 0.72, 95%CI (0.71–0.73)). Higher age, BMI and prescription of non-steroidal anti-inflammatory drugs contributed most to increased OA risk prediction ahead of diagnosis. We identified 14 subgroups of OA risk profiles. These subgroups were validated in an independent set of patients evaluating the 11-year OA risk, with 88% of patients being uniquely assigned to one of the 14 subgroups. Individual OA risk profiles were characterised by personalised biomarkers. Omics integration demonstrated the predictive importance of key OA genes and pathways (e.g., GDF5 and TGF-β signalling) and OA-specific biomarkers (e.g., CRTAC1 and COL9A1). In summary, this work identifies opportunities for personalised OA prevention and insights into its underlying pathogenesis. Osteoarthritis can be caused by multiple biological mechanisms but the drivers of disease risk are not well understood. Here, the authors use data from UK Biobank in machine learning models to identify clinical and biological markers associated with development of osteoarthritis and identify sub-groups with different risk profiles. [ABSTRACT FROM AUTHOR]

Left ventricular regional wall thickness (LVRWT) is an independent predictor of morbidity and mortality in cardiovascular diseases (CVDs). To identify specific genetic influences on individual LVRWT, we established a novel deep learning algorithm to calculate 12 LVRWTs accurately in 42,194 individuals from the UK Biobank with cardiac magnetic resonance (CMR) imaging. Genome-wide association studies of CMR-derived 12 LVRWTs identified 72 significant genetic loci associated with at least one LVRWT phenotype (P < 5 × 10−8), which were revealed to actively participate in heart development and contraction pathways. Significant causal relationships were observed between the LVRWT traits and hypertrophic cardiomyopathy (HCM) using genetic correlation and Mendelian randomization analyses (P < 0.01). The polygenic risk score of inferoseptal LVRWT at end systole exhibited a notable association with incident HCM, facilitating the identification of high-risk individuals. The findings yield insights into the genetic determinants of LVRWT phenotypes and shed light on the biological basis for HCM etiology. Changes of left ventricular structure are used to predict morbidity and mortality in cardiovascular diseases. Here the authors conducted a study using advanced deep learning technology to analyze left ventricular regional wall thickness (LVRWT) in a large population, identifying 72 significant genetic loci linked to LVRWT traits. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic led to unparalleled pressure on healthcare services. Improved healthcare planning in relation to diseases affecting the respiratory system has consequently become a key concern. We investigated the value of integrating sales of non-prescription medications commonly bought for managing respiratory symptoms, to improve forecasting of weekly registered deaths from respiratory disease at local levels across England, by using over 2 billion transactions logged by a UK high street retailer from March 2016 to March 2020. We report the results from the novel AI (Artificial Intelligence) explainability variable importance tool Model Class Reliance implemented on the PADRUS model (Prediction of Amount of Deaths by Respiratory disease Using Sales). PADRUS is a machine learning model optimised to predict registered deaths from respiratory disease in 314 local authority areas across England through the integration of shopping sales data and focused on purchases of non-prescription medications. We found strong evidence that models incorporating sales data significantly out-perform other models that solely use variables traditionally associated with respiratory disease (e.g. sociodemographics and weather data). Accuracy gains are highest (increases in R2 (coefficient of determination) between 0.09 to 0.11) in periods of maximum risk to the general public. Results demonstrate the potential to utilise sales data to monitor population health with information at a high level of geographic granularity. Novel indicators of infectious disease prevalence could improve real-time surveillance and support healthcare planning. Here, the authors show that sales data for non-prescription medications from a UK high street retailer can improve the accuracy of models forecasting mortality from respiratory infections. [ABSTRACT FROM AUTHOR]

The emergency department (ED) is a fast-paced environment responsible for large volumes of patients with varied disease acuity. Operational pressures on EDs are increasing, which creates the imperative to efficiently identify patients at imminent risk of acute deterioration. The aim of this study is to systematically compare the performance of machine learning algorithms based on logistic regression, gradient boosted decision trees, and support vector machines for predicting imminent clinical deterioration for patients based on cross-sectional patient data extracted from electronic patient records (EPR) at the point of entry to the hospital. We apply state-of-the-art machine learning methods to predict early patient deterioration, based on their first recorded vital signs, observations, laboratory results, and other predictors documented in the EPR. Clinical deterioration in this study is measured by in-hospital mortality and/or admission to critical care. We build on prior work by incorporating interpretable machine learning and fairness-aware modelling, and use a dataset comprising 118, 886 unplanned admissions to Salford Royal Hospital, UK, to systematically compare model variations for predicting mortality and critical care utilisation within 24 hours of admission. We compare model performance to the National Early Warning Score 2 (NEWS2) and yield up to a 0.366 increase in average precision, up to a 21.16 % reduction in daily alert rate, and a median 0.599 reduction in differential bias amplification across the protected demographics of age and sex. We use Shapely Additive exPlanations to justify the models' outputs, verify that the captured data associations align with domain knowledge, and pair predictions with the causal context of each patient's most influential characteristics. Introducing our modelling to clinical practice has the potential to reduce alert fatigue and identify high-risk patients with a lower NEWS2 that might be missed currently, but further work is needed to trial the models in clinical practice. We encourage future research to follow a systematised approach to data-driven risk modelling to obtain clinically applicable support tools. [ABSTRACT FROM AUTHOR]

Mental wellbeing of university students is a growing concern that has been worsening during the COVID-19 pandemic. Numerous studies have gathered empirical data to explore the mental health impact of the pandemic on university students and investigate factors associated with higher levels of distress. While the online questionnaire survey has been a prevalent means to collect data, regression analysis has been observed a dominating approach to interpret and understand the impact of independent factors on a mental wellbeing state of interest. Drawbacks such as sensitivity to outliers, ineffectiveness in case of multiple predictors highly correlated may limit the use of regression in complex scenarios. These observations motivate the underlying research to propose alternative computational methods to investigate the questionnaire data. Inspired by recent machine learning advances, this research aims to construct a framework through feature permutation importance to empower the application of a variety of machine learning algorithms that originate from different computational frameworks and learning theories, including algorithms that cannot directly provide exact numerical contributions of individual factors. This would enable to explore quantitative impact of predictors in influencing student mental wellbeing from multiple perspectives as a result of using different algorithms, thus complementing the single view due to the dominant use of regression. Applying the proposed approach over an online survey in a UK university, the analysis suggests the past medical record and wellbeing history and the experience of adversity contribute significantly to mental wellbeing states; and the frequent communication with families and friends to keep good relationship as well as regular exercise are generally contributing to improved mental wellbeing. [ABSTRACT FROM AUTHOR]

Cardiovascular ageing is a process that begins early in life and leads to a progressive change in structure and decline in function due to accumulated damage across diverse cell types, tissues and organs contributing to multi-morbidity. Damaging biophysical, metabolic and immunological factors exceed endogenous repair mechanisms resulting in a pro-fibrotic state, cellular senescence and end-organ damage, however the genetic architecture of cardiovascular ageing is not known. Here we use machine learning approaches to quantify cardiovascular age from image-derived traits of vascular function, cardiac motion and myocardial fibrosis, as well as conduction traits from electrocardiograms, in 39,559 participants of UK Biobank. Cardiovascular ageing is found to be significantly associated with common or rare variants in genes regulating sarcomere homeostasis, myocardial immunomodulation, and tissue responses to biophysical stress. Ageing is accelerated by cardiometabolic risk factors and we also identify prescribed medications that are potential modifiers of ageing. Through large-scale modelling of ageing across multiple traits our results reveal insights into the mechanisms driving premature cardiovascular ageing and reveal potential molecular targets to attenuate age-related processes. Cardiovascular ageing is characterised by a progressive decline in function, which contributes to multi-morbidity. Here, the authors use machine learning to predict biological age and identify key genetic risk factors. [ABSTRACT FROM AUTHOR]

We aimed to identify potential novel predictors for breast cancer among post-menopausal women, with pre-specified interest in the role of polygenic risk scores (PRS) for risk prediction. We utilised an analysis pipeline where machine learning was used for feature selection, prior to risk prediction by classical statistical models. An "extreme gradient boosting" (XGBoost) machine with Shapley feature-importance measures were used for feature selection among ≈ 1.7 k features in 104,313 post-menopausal women from the UK Biobank. We constructed and compared the "augmented" Cox model (incorporating the two PRS, known and novel predictors) with a "baseline" Cox model (incorporating the two PRS and known predictors) for risk prediction. Both of the two PRS were significant in the augmented Cox model ( p < 0.001 ). XGBoost identified 10 novel features, among which five showed significant associations with post-menopausal breast cancer: plasma urea (HR = 0.95, 95% CI 0.92–0.98, p < 0.001 ), plasma phosphate (HR = 0.68, 95% CI 0.53–0.88, p = 0.003 ), basal metabolic rate (HR = 1.17, 95% CI 1.11–1.24, p < 0.001 ), red blood cell count (HR = 1.21, 95% CI 1.08–1.35, p < 0.001 ), and creatinine in urine (HR = 1.05, 95% CI 1.01–1.09, p = 0.006 ). Risk discrimination was maintained in the augmented Cox model, yielding C-index 0.673 vs 0.667 (baseline Cox model) with the training data and 0.665 vs 0.664 with the test data. We identified blood/urine biomarkers as potential novel predictors for post-menopausal breast cancer. Our findings provide new insights to breast cancer risk. Future research should validate novel predictors, investigate using multiple PRS and more precise anthropometry measures for better breast cancer risk prediction. [ABSTRACT FROM AUTHOR]

Diffuse idiopathic skeletal hyperostosis (DISH) is a condition where adjacent vertebrae become fused through formation of osteophytes. The genetic and epidemiological etiology of this condition is not well understood. Here, we implemented a machine learning algorithm to assess the prevalence and severity of the pathology in ~40,000 lateral DXA scans in the UK Biobank Imaging cohort. We find that DISH is highly prevalent, above the age of 45, ~20% of men and ~8% of women having multiple osteophytes. Surprisingly, we find strong phenotypic and genetic association of DISH with increased bone mineral density and content throughout the entire skeletal system. Genetic association analysis identified ten loci associated with DISH, including multiple genes involved in bone remodeling (RUNX2, IL11, GDF5, CCDC91, NOG, and ROR2). Overall, this study describes genetics of DISH and implicates the role of overactive osteogenesis as a key driver of the pathology. The genetic contribution to risk of Diffuse Idiopathic Skeletal Hyperostosis has been unclear. Here, the authors find genetic variation associated with Diffuse Idiopathic Skeletal Hyperostosis, finding phenotypic and genetic association with increased bone mass throughout the skeletal system. [ABSTRACT FROM AUTHOR]

Distributed lags play important roles in explaining the short-run dynamic and long-run cumulative effects of features on a response variable. Unlike the usual lag length selection, important lags with significant weights are selected in a distributed lag model (DLM). Inspired by the importance of distributed lags, this research focuses on the construction of distributed lag inspired machine learning (DLIML) for predicting vaccine-induced changes in COVID-19 hospitalization and intensive care unit (ICU) admission rates. Importance of a lagged feature in DLM is examined by hypothesis testing and a subset of important features are selected by evaluating an information criterion. Akin to the DLM, we demonstrate the selection of distributed lags in machine learning by evaluating importance scores and objective functions. Finally, we apply the DLIML with supervised learning for forecasting daily changes in COVID-19 hospitalization and ICU admission rates in United Kingdom (UK) and United States of America (USA). A sharp decline in hospitalization and ICU admission rates are observed when around 40% people are vaccinated. For one percent more vaccination, daily changes in hospitalization and ICU admission rates are expected to reduce by 4.05 and 0.74 per million after 14 days in UK, and 5.98 and 1.04 per million after 20 days in USA, respectively. Long-run cumulative effects in the DLM demonstrate that the daily changes in hospitalization and ICU admission rates are expected to jitter around the zero line in a long-run. Application of the DLIML selects fewer lagged features but provides qualitatively better forecasting outcome for data-driven healthcare service planning. [ABSTRACT FROM AUTHOR]

Machine learning for hospital operations is under-studied. We present a prediction pipeline that uses live electronic health-records for patients in a UK teaching hospital's emergency department (ED) to generate short-term, probabilistic forecasts of emergency admissions. A set of XGBoost classifiers applied to 109,465 ED visits yielded AUROCs from 0.82 to 0.90 depending on elapsed visit-time at the point of prediction. Patient-level probabilities of admission were aggregated to forecast the number of admissions among current ED patients and, incorporating patients yet to arrive, total emergency admissions within specified time-windows. The pipeline gave a mean absolute error (MAE) of 4.0 admissions (mean percentage error of 17%) versus 6.5 (32%) for a benchmark metric. Models developed with 104,504 later visits during the Covid-19 pandemic gave AUROCs of 0.68–0.90 and MAE of 4.2 (30%) versus a 4.9 (33%) benchmark. We discuss how we surmounted challenges of designing and implementing models for real-time use, including temporal framing, data preparation, and changing operational conditions. [ABSTRACT FROM AUTHOR]

As patient health information is highly regulated due to privacy concerns, most machine learning (ML)-based healthcare studies are unable to test on external patient cohorts, resulting in a gap between locally reported model performance and cross-site generalizability. Different approaches have been introduced for developing models across multiple clinical sites, however less attention has been given to adopting ready-made models in new settings. We introduce three methods to do this—(1) applying a ready-made model "as-is" (2); readjusting the decision threshold on the model's output using site-specific data and (3); finetuning the model using site-specific data via transfer learning. Using a case study of COVID-19 diagnosis across four NHS Hospital Trusts, we show that all methods achieve clinically-effective performances (NPV > 0.959), with transfer learning achieving the best results (mean AUROCs between 0.870 and 0.925). Our models demonstrate that site-specific customization improves predictive performance when compared to other ready-made approaches. [ABSTRACT FROM AUTHOR]

We present a simple and efficient hypothesis-free machine learning pipeline for risk factor discovery that accounts for non-linearity and interaction in large biomedical databases with minimal variable pre-processing. In this study, mortality models were built using gradient boosting decision trees (GBDT) and important predictors were identified using a Shapley values-based feature attribution method, SHAP values. Cox models controlled for false discovery rate were used for confounder adjustment, interpretability, and further validation. The pipeline was tested using information from 502,506 UK Biobank participants, aged 37–73 years at recruitment and followed over seven years for mortality registrations. From the 11,639 predictors included in GBDT, 193 potential risk factors had SHAP values ≥ 0.05, passed the correlation test, and were selected for further modelling. Of the total variable importance summed up, 60% was directly health related, and baseline characteristics, sociodemographics, and lifestyle factors each contributed about 10%. Cox models adjusted for baseline characteristics, showed evidence for an association with mortality for 166 out of the 193 predictors. These included mostly well-known risk factors (e.g., age, sex, ethnicity, education, material deprivation, smoking, physical activity, self-rated health, BMI, and many disease outcomes). For 19 predictors we saw evidence for an association in the unadjusted but not adjusted analyses, suggesting bias by confounding. Our GBDT-SHAP pipeline was able to identify relevant predictors 'hidden' within thousands of variables, providing an efficient and pragmatic solution for the first stage of hypothesis free risk factor identification. [ABSTRACT FROM AUTHOR]