Precision global health for real-time action

Precision global health, augmented with artificial intelligence, has the potential to address transnational problems (eg, outbreaks of emerging infectious diseases, diabetes, addictions, ageing, or mental health) and deliver targeted and effective interventions through integrated approaches which combine life sciences, social sciences, and data sciences with public support. More than half the global population are connected to the Internet, mainly through mobile phones, and several countries in sub-Saharan Africa are leading the annual growth of active mobile social users with over 17% in 2018. Thus, the role of local populations and civil society is more important than ever, to identify challenges and work together to address some of the most pressing global health issues and sustainable development.

The full commentary can be downloaded below.  

Prescription opioid misuse motive latent classes: outcomes from a nationally representative US sample

Prescription opioid misuse (POM) contributes to a larger opioid crisis in the US and Canada, with over 17 000 US POM-related overdose deaths in 2017. Our aims were to (1) identify specific profiles of respondents based on POM motives using the US National Survey on Drug Use and Health (NSDUH) and (2) compare profile respondents on sociodemographics, substance use and mental and physical health outcomes.

Analyses included 2017–18 NSDUH respondents with data on POM motives (n = 4810). POM was defined as prescription opioid use in a way not intended by the prescriber, including use without a prescription, in larger amounts or more frequently. Nine POM motives for the most recent episode were assessed, including ‘to relieve physical pain’ and ‘to get high’. Latent classes, based on POM motives, were estimated. Classes were compared on sociodemographics, substance use and physical and mental health outcomes.

Eight latent classes were identified (in order of prevalence): pain relief only, relaxpain relief, sleep-pain relief, multi-motive, high, experimenter, emotional coping and dependent/hooked. Compared to the pain relief only group, the high and multi-motive classes had higher odds of all substance use outcomes, with the dependent/hooked class having higher odds on all but one outcome. Six of the eight classes had higher odds of past-year mental health treatment and suicidal ideation than the pain relief only class.

Screening for pain, pain conditions, problematic substance use and psychopathology are recommended in those with any POM. While those in the dependent/hooked, multi-motive and emotional coping classes are most likely to have prescription opioid use disorder (OUD), screening for OUD symptoms in all individuals with POM is also warranted.

The full article can be downloaded below.

Combining deep learning with token selection for patient phenotyping from electronic health records

Artificial intelligence provides the opportunity to reveal important information buried in large amounts of complex data. Electronic health records (eHRs) are a source of such big data that provide a multitude of health related clinical information about patients. However, text data from eHRs, e.g., discharge summary notes, are challenging in their analysis because these notes are free-form texts and the writing formats and styles vary considerably between different records. For this reason, in this paper we study deep learning neural networks in combination with natural language processing to analyze text data from clinical discharge summaries. We provide a detail analysis of patient phenotyping, i.e., the automatic prediction of ten patient disorders, by investigating the influence of network architectures, sample sizes and information content of tokens. Importantly, for patients suffering from Chronic Pain, the disorder that is the most diffcult one to classify, we fnd the largest performance gain for a combined word- and sentence-level input convolutional neural network (ws-CNN). As a general result, we fnd that the combination of data quality and data quantity of the text data is playing a crucial role for using more complex network architectures that improve significantly beyond a word-level input CNN model. From our investigations of learning curves and token selection mechanisms, we conclude that for such a transition one requires larger sample sizes because the amount of information per sample is quite small and only carried by few tokens and token categories. Interestingly, we found that the token frequency in the eHRs follow a Zipf law and we utilized this behavior to investigate the information content of tokens by defining a token selection mechanism. The latter addresses also issues of explainable AI.

The full article can be downloaded below.  

U.S. Health Care from a Global Perspective, 2019: Higher Spending, Worse Outcomes?

This analysis is the latest in a series of Commonwealth Fund cross-national comparisons that uses health data from the Organisation for Economic Co-operation and Development (OECD) to assess U.S. health care system spending, outcomes, risk factors and prevention, utilization, and quality, relative to 10 other high-income countries: Australia, Canada, France, Germany, the Netherlands, New Zealand, Norway, Sweden, Switzerland, and the United Kingdom. We also compare U.S. performance to that of the OECD average, comprising 36 high-income member countries.

The full Commonwealth Fund analysis can be viewed at this link.  

This case study highlights the success of a cross-specialty diabetes pilot program, implemented by Geisinger Health System (Geisinger), to help control type 2 diabetes amongst Keystone Accountable Care Organization (ACO) beneficiaries. Geisinger, a member of the Keystone ACO, began to implement the program in early 2019 across Geisinger primary care providers (PCP). Beneficiaries continue to see reductions in their A1C levels, and the program will expand to other chronic conditions and be rolled out to providers throughout the entire ACO network.

In June of 2019, the Geisinger Health Plan’s type 2 diabetes prevention program earned full recognition from the Centers for Disease Control and Prevention (CDC), making it one of only 20 program suppliers in Pennsylvania to do so. Full recognition status is reserved for programs that effectively deliver high-quality, evidence-based programming through maintaining participant retention and showing clinical outcomes that meet all the standards for CDC recognition. Using interviews with the Geisinger team, this case study details how Geisinger continued its success managing type 2 diabetes within its ACO population.

How AI is battling the coronavirus outbreak

When a mysterious illness first pops up, it can be difficult for governments and public health officials to gather information quickly and coordinate a response. But new artificial intelligence technology can automatically mine through news reports and online content from around the world, helping experts recognize anomalies that could lead to a potential epidemic or, worse, a pandemic. In other words, our new AI overlords might actually help us survive the next plague.

These new AI capabilities are on full display with the recent outbreak of COVID-19, which was identified early by a Canadian firm called BlueDot, which is one of a number of companies that use data to evaluate public health risks. The company, which says it conducts “automated infectious disease surveillance,” notified its customers about the new form of coronavirus at the end of December, days before both the US Centers for Disease Control and Prevention (CDC) and the World Health Organization (WHO) sent out official notices, as reported by Wired. Now nearing the end of January, the respiratory virus that’s been linked to the city of Wuhan in China has already claimed the lives of more than 100 people. Cases have also popped up in several other countries, including the United States, and the CDC is warning Americans to avoid non-essential travel to China.

The full Vox article can be viewed at this link.  

AI: Pharma's Perfect Medicine

AI appears set to be the thing that separates the next generation of business success stories and market dropouts. It has revolutionized the transportation industry by bringing the science fiction dream of autonomous cars into reality, as driverless taxis have already been tested and deployed in the U.S.

Further indicators of its importance come from finance companies like Goldman Sachs, JPMorgan and Morgan Stanley — all of which have aggressively expanded their data and tech teams over the past year — looking to deploy AI projects that will give them the competitive edge against their rivals. The application of this technology ranges from the mundane to the absurd, seemingly with no sector able to escape its influence — and the pharma industry is no different.

The full Forbes article can be viewed at this link.  

When Insurance Won't Cover Drugs, Americans Make 'Tough Choices' About Their Health

The majority of Americans have health insurance that includes coverage for prescription drugs. But unfortunately that doesn't ensure that they can afford the specific drugs their doctors prescribe for them.

In fact, many Americans report that their insurance plans sometimes don't cover a drug they need — and nearly half the people whom this happens to say they simply don't fill the prescription. That's according to a poll released this month on income inequality from NPR, the Robert Wood Johnson Foundation and the Harvard T.H. Chan School of Public Health.

The full NPR article can be viewed at this link.  

LIFE EXPERIENCES AND INCOME INEQUALITY IN THE UNITED STATES

Over the past five decades, income inequality has sharply increased between the highest income earners and middle- and lower-income earners in the U.S. This poll examines the implications of this growing inequality for the lives of U.S. adults across different income levels. This report, Life Experiences and Income Inequality in the United States, is based on a survey conducted for NPR, the Robert Wood Johnson Foundation, and the Harvard T.H. Chan School of Public Health. It explores how adults’ personal experiences, values, and views on income inequality, opportunity, and health in the U.S., differ among adults by household income. Surveyed adults are split into four income categories: those in the top 1% highest income households in the U.S. (earning at least $500,000/year), those in higher-income households (earning $100,000-$499,999/year), those in middle-income households (earning $35,000-$99,999/year), and those in lower-income households (earning less than $35,000/year). Due to the heterogeneity of incomes in the higher-income category, analyses in this report focus on differences between the top 1% highest income adults compared to middle- and lower-income adults, though results are included for all four income groups. It was conducted July 17 – August 18, 2019, among a nationally representative, probability-based telephone (cell and landline) sample of 1,885 adults ages 18 or older living in the United States.

The full report can be downloaded below.