Unsupervised machine learning predicts future sexual behaviour and sexually transmitted infections among HIV-positive men who have sex with men.

Machine learning is increasingly introduced into medical fields, yet there is limited evidence for its benefit over more commonly used statistical methods in epidemiological studies. We introduce an unsupervised machine learning framework for longitudinal features and evaluate it using sexual behaviour data from the last 20 years from over 3'700 participants in the Swiss HIV Cohort Study (SHCS). We use hierarchical clustering to find subgroups of men who have sex with men in the SHCS with similar sexual behaviour up to May 2017, and apply regression to test whether these clusters enhance predictions of sexual behaviour or sexually transmitted diseases (STIs) after May 2017 beyond what can be predicted with conventional parameters. We find that behavioural clusters enhance model performance according to likelihood ratio test, Akaike information criterion and area under the receiver operator characteristic curve for all outcomes studied, and according to Bayesian information criterion for five out of ten outcomes, with particularly good performance for predicting future sexual behaviour and recurrent STIs. We thus assess a methodology that can be used as an alternative means for creating exposure categories from longitudinal data in epidemiological models, and can contribute to the understanding of time-varying risk factors.

HIV testing history and access to treatment among migrants living with HIV in Europe.

INTRODUCTION: Migrants are overrepresented in the European HIV epidemic. We aimed to understand the barriers and facilitators to HIV testing and current treatment and healthcare needs of migrants living with HIV in Europe. METHODS: A cross-sectional study was conducted in 57 HIV clinics in nine countries (Belgium, Germany, Greece, Italy, The Netherlands, Portugal, Spain, Switzerland and United Kingdom), July 2013 to July 2015. HIV-positive patients were eligible for inclusion if they were as follows: 18 years or older; foreign-born residents and diagnosed within five years of recruitment. Questionnaires were completed electronically in one of 15 languages and linked to clinical records. Primary outcomes were access to primary care and previous negative HIV test. Data were analysed using random effects logistic regression. Outcomes of interest are presented for women, heterosexual men and gay/bisexual men. RESULTS: A total of 2093 respondents (658 women, 446 heterosexual men and 989 gay/bisexual men) were included. The prevalence of a previous negative HIV test was 46.7%, 43.4% and 82.0% for women, heterosexual and gay/bisexual men respectively. In multivariable analysis previous testing was positively associated with: receipt of post-migration antenatal care among women, permanent residency among heterosexual men and identifying as gay rather than bisexual among gay/bisexual men. Access to primary care was found to be high (>83%) in all groups and was strongly associated with country of residence. Late diagnosis was common for women and heterosexual men (60.8% and 67.1%, respectively) despite utilization of health services prior to diagnosis. Across all groups almost three-quarters of people on antiretrovirals had an HIV viral load <50 copies/mL. CONCLUSIONS: Migrants access healthcare in Europe and while many migrants had previously tested for HIV, that they went on to test positive at a later date suggests that opportunities for HIV prevention are being missed. Expansion of testing beyond sexual health and antenatal settings is still required and testing opportunities should be linked with combination prevention measures such as access to PrEP and treatment as prevention.

Late presentation to HIV care despite good access to health services: current epidemiological trends and how to do better.

In 2014, there were 36.9 million people worldwide living with human immunodeficiency virus (PLWH), of whom 17.1 million did not know they were infected. Whilst the number of new human immunodeficiency virus (HIV) infections has declined globally since 2000, there are still regions where new infection rates are rising, and diagnosing HIV early in the course of infection remains a challenge. Late presentation to care in HIV refers to individuals newly presenting for HIV care with a CD4 count below 350 cells/µl or with an acquired immune deficiency syndrome (AIDS)-defining event. Late presentation is associated with increased patient morbidity and mortality, healthcare costs and risk of onward transmission by individuals unaware of their status. Further, late presentation limits the effectiveness of all subsequent steps in the cascade of HIV care. Recent figures from 34 countries in Europe show that late presentation occurs in 38.3% to 49.8% of patients newly presenting for care, depending on region. In Switzerland, data from patients enrolled in the Swiss HIV Cohort Study put the rate of late presentation at 49.8% and show that patients outside established HIV risk groups are most likely to be late presenters. Provider-initiated testing needs to be improved to reach these groups, which include heterosexual men and women and older patients. The aim of this review is to describe the scale and implications of late presentation using cohort data from Switzerland and elsewhere in Europe, and to highlight initiatives to improve early HIV diagnosis. The importance of recognising indicator conditions and the potential for missed opportunities for HIV testing is illustrated in three clinical case studies.