Assessing machine learning for fair prediction of ADHD in school pupils using a retrospective cohort study of linked education and healthcare data.

OBJECTIVES: Attention deficit hyperactivity disorder (ADHD) is a prevalent childhood disorder, but often goes unrecognised and untreated. To improve access to services, accurate predictions of populations at high risk of ADHD are needed for effective resource allocation. Using a unique linked health and education data resource, we examined how machine learning (ML) approaches can predict risk of ADHD. DESIGN: Retrospective population cohort study. SETTING: South London (2007-2013). PARTICIPANTS: n=56 258 pupils with linked education and health data. PRIMARY OUTCOME MEASURES: Using area under the curve (AUC), we compared the predictive accuracy of four ML models and one neural network for ADHD diagnosis. Ethnic group and language biases were weighted using a fair pre-processing algorithm. RESULTS: Random forest and logistic regression prediction models provided the highest predictive accuracy for ADHD in population samples (AUC 0.86 and 0.86, respectively) and clinical samples (AUC 0.72 and 0.70). Precision-recall curve analyses were less favourable. Sociodemographic biases were effectively reduced by a fair pre-processing algorithm without loss of accuracy. CONCLUSIONS: ML approaches using linked routinely collected education and health data offer accurate, low-cost and scalable prediction models of ADHD. These approaches could help identify areas of need and inform resource allocation. Introducing 'fairness weighting' attenuates some sociodemographic biases which would otherwise underestimate ADHD risk within minority groups.

Impact of COVID-19 on healthcare organisation and cancer outcomes.

BACKGROUND: Changes in the management of patients with cancer and delays in treatment delivery during the COVID-19 pandemic may impact the use of hospital resources and cancer mortality. PATIENTS AND METHODS: Patient flows, patient pathways and use of hospital resources during the pandemic were simulated using a discrete event simulation model and patient-level data from a large French comprehensive cancer centre's discharge database, considering two scenarios of delays: massive return of patients from November 2020 (early-return) or March 2021 (late-return). Expected additional cancer deaths at 5 years and mortality rate were estimated using individual hazard ratios based on literature. RESULTS: The number of patients requiring hospital care during the simulation period was 13,000. In both scenarios, 6-8% of patients were estimated to present a delay of >2 months. The overall additional cancer deaths at 5 years were estimated at 88 in early-return and 145 in late-return scenario, with increased additional deaths estimated for sarcomas, gynaecological, liver, head and neck, breast cancer and acute leukaemia. This represents a relative additional cancer mortality rate at 5 years of 4.4 and 6.8% for patients expected in year 2020, 0.5 and 1.3% in 2021 and 0.5 and 0.5% in 2022 for each scenario, respectively. CONCLUSIONS: Pandemic-related diagnostic and treatment delays in patients with cancer are expected to impact patient survival. In the perspective of recurrent pandemics or alternative events requiring an intensive use of limited hospital resources, patients should be informed not to postpone care, and medical resources for patients with cancer should be sanctuarised.

Patterns of use of the Mental Health Act 1983, from 2007-2008 to 2016-2017, in two major London secondary mental healthcare providers.

Trends in detention under the Mental Health Act 1983 in two major London secondary mental healthcare providers were explored using patient-level data in a historical cohort study between 2007-2008 and 2016-2017. An increase in the number of detention episodes initiated per fiscal year was observed at both sites. The rise was accompanied by an increase in the number of active patients; the proportion of active patients detained per year remained relatively stable. Findings suggest that the rise in the number of detentions reflects the rise of the number of people receiving secondary mental healthcare.