Web search activity data accurately predict population chronic disease risk in the USA

BACKGROUND: The WHO framework for non-communicable disease (NCD) describes risks and outcomes comprising the majority of the global burden of disease. These factors are complex and interact at biological, behavioural, environmental and policy levels presenting challenges for population monitoring and intervention evaluation. This paper explores the utility of machine learning methods applied to population-level web search activity behaviour as a proxy for chronic disease risk factors. METHODS: Web activity output for each element of the WHO's Causes of NCD framework was used as a basis for identifying relevant web search activity from 2004 to 2013 for the USA. Multiple linear regression models with regularisation were used to generate predictive algorithms, mapping web search activity to Centers for Disease Control and Prevention (CDC) measured risk factor/disease prevalence. Predictions for subsequent target years not included in the model derivation were tested against CDC data from population surveys using Pearson correlation and Spearman's r. RESULTS: For 2011 and 2012, predicted prevalence was very strongly correlated with measured risk data ranging from fruits and vegetables consumed (r=0.81; 95% CI 0.68 to 0.89) to alcohol consumption (r=0.96; 95% CI 0.93 to 0.98). Mean difference between predicted and measured differences by State ranged from 0.03 to 2.16. Spearman's r for state-wise predicted versus measured prevalence varied from 0.82 to 0.93. CONCLUSIONS: The high predictive validity of web search activity for NCD risk has potential to provide real-time information on population risk during policy implementation and other population-level NCD prevention efforts.

Early protection in sheep against intratypic heterologous challenge with serotype O foot-and-mouth disease virus using high-potency, emergency vaccine

In 2009-2011, spread of a serotype O foot-and-mouth disease virus (FMDV) belonging to the South East Asia topotype led to the culling of over 3.5 million cattle and pigs in Japan and Korea. The O1 Manisa vaccine (belonging to the Middle East-South Asian topotype) was used at high potency in Korea to limit the expansion of the outbreak. However, no data are available on the spread of this virus or the efficacy of the O1 Manisa vaccine against this virus in sheep. In this study, the early protection afforded with a high potency (>6 PD50) FMD O1 Manisa vaccine against challenge with the O/SKR/2010 virus was tested in sheep. Sheep (n=8) were vaccinated 4 days prior to continuous direct-contact challenge with donor sheep. Donor sheep were infected with FMDV O/SKR/2010 by coronary band inoculation 24h prior to contact with the vaccinated animals, or unvaccinated controls (n=4). Three of the four control sheep became infected, two clinically. All eight O1 Manisa vaccinated sheep were protected from clinical disease. None had detectable antibodies to FMDV non-structural proteins (3ABC), no virus was isolated from nasal swabs, saliva or oro-pharyngeal fluid and none became carriers. Using this model of challenge, sheep were protected against infection as early as 4 days post vaccination.

Evaluation of a knowledge translation and exchange platform to advance non-communicable disease prevention

Coordinated systems are required to ensure evidence-informed practice and evaluation of community-based interventions (CBIs). Knowledge translation and exchange (KTE) strategies show promise, but these require evaluation. This paper describes implementation and evaluation of COOPS, a national KTE platform to support best practice in obesity prevention CBIs. A logic model guides KTE activities including knowledge brokering, networking, tailored communications, training, and needs assessments. A mixed-methods evaluation includes communications data, knowledge brokering database, annual survey of CBIs, pre- and post-event questionnaires, interviews, social network analysis, and case studies. This evaluation will contribute to understanding the process of implementing a KTE platform with CBIs and its reach, quality and effectiveness.

Health improvement and prevention study (HIPS) - evaluation of an intervention to prevent vascular disease in general practice

Background

The Health Improvement and Prevention Study (HIPS) study aims to evaluate the capacity of general practice to identify patients at high risk for developing vascular disease and to reduce their risk of vascular disease and diabetes through behavioural interventions delivered in general practice and by the local primary care organization.

Methods/Design

HIPS is a stratified randomized controlled trial involving 30 general practices in NSW, Australia. Practices are randomly allocated to an 'intervention' or 'control' group. General practitioners (GPs) and practice nurses (PNs) are offered training in lifestyle counselling and motivational interviewing as well as practice visits and patient educational resources. Patients enrolled in the trial present for a health check in which the GP and PN provide brief lifestyle counselling based on the 5As model (ask, assess, advise, assist, and arrange) and refer high risk patients to a diet education and physical activity program. The program consists of two individual visits with a dietician or exercise physiologist and four group sessions, after which patients are followed up by the GP or PN. In each practice 160 eligible patients aged between 40 and 64 years are invited to participate in the study, with the expectation that 40 will be eligible and willing to participate. Evaluation data collection consists of (1) a practice questionnaire, (2) GP and PN questionnaires to assess preventive care attitudes and practices, (3) patient questionnaire to assess self-reported lifestyle behaviours and readiness to change, (4) physical assessment including weight, height, body mass index (BMI), waist circumference and blood pressure, (5) a fasting blood test for glucose and lipids, (6) a clinical record audit, and (7) qualitative data collection. All measures are collected at baseline and 12 months except the patient questionnaire which is also collected at 6 months. Study outcomes before and after the intervention is compared between intervention and control groups after adjusting for baseline differences and clustering at the level of the practice.

Discussion

This study will provide evidence of the effectiveness of a primary care intervention to reduce the risk of cardiovascular disease and diabetes in general practice patients. It will inform current policies and programs designed to prevent these conditions in Australian primary health care.

DeepCare: a deep dynamic memory model for predictive medicine

Personalized predictive medicine necessitates the modeling of patient illness and care processes, which inherently have long-term temporal dependencies. Healthcare observations, recorded in electronic medical records, are episodic and irregular in time. We introduce DeepCare, an end-to-end deep dynamic neural network that reads medical records, stores previous illness history, infers current illness states and predicts future medical outcomes. At the data level, DeepCare represents care episodes as vectors in space, models patient health state trajectories through explicit memory of historical records. Built on Long Short-Term Memory (LSTM), DeepCare introduces time parameterizations to handle irregular timed events by moderating the forgetting and consolidation of memory cells. DeepCare also incorporates medical interventions that change the course of illness and shape future medical risk. Moving up to the health state level, historical and present health states are then aggregated through multiscale temporal pooling, before passing through a neural network that estimates future outcomes. We demonstrate the efficacy of DeepCare for disease progression modeling, intervention recommendation, and future risk prediction. On two important cohorts with heavy social and economic burden -- diabetes and mental health -- the results show improved modeling and risk prediction accuracy.