Fulfilling the Promise of Artificial Intelligence in the Health Sector: Let's Get Real.

OBJECTIVES: This study aimed to showcase the potential and key concerns and risks of artificial intelligence (AI) in the health sector, illustrating its application with current examples, and to provide policy guidance for the development, assessment, and adoption of AI technologies to advance policy objectives. METHODS: Nonsystematic scan and analysis of peer-reviewed and gray literature on AI in the health sector, focusing on key insights for policy and governance. RESULTS: The application of AI in the health sector is currently in the early stages. Most applications have not been scaled beyond the research setting. The use in real-world clinical settings is especially nascent, with more evidence in public health, biomedical research, and "back office" administration. Deploying AI in the health sector carries risks and hazards that must be managed proactively by policy makers. For AI to produce positive health and policy outcomes, 5 key areas for policy are proposed, including health data governance, operationalizing AI principles, flexible regulation, skills among health workers and patients, and strategic public investment. CONCLUSIONS: AI is not a panacea, but a tool to address specific problems. Its successful development and adoption require data governance that ensures high-quality data are available and secure; relevant actors can access technical infrastructure and resources; regulatory frameworks promote trustworthy AI products; and health workers and patients have the information and skills to use AI products and services safely, effectively, and efficiently. All of this requires considerable investment and international collaboration.

Assessing data protection and governance in health information systems: a novel methodology of Privacy and Ethics Impact and Performance Assessment (PEIPA).

BACKGROUND: Data processing of health research databases often requires a Data Protection Impact Assessment to evaluate the severity of the risk and the appropriateness of measures taken to comply with the European Union (EU) General Data Protection Regulation (GDPR). We aimed to define and apply a comprehensive method for the evaluation of privacy, data governance and ethics among research networks involved in the EU Project Bridge Health. METHODS: Computerised survey among associated partners of main EU Consortia, using a targeted instrument designed by the principal investigator and progressively refined in collaboration with an international advisory panel. Descriptive measures using the percentage of adoption of privacy, data governance and ethical principles as main endpoints were used for the analysis and interpretation of the results. RESULTS: A total of 15 centres provided relevant information on the processing of sensitive data from 10 European countries. Major areas of concern were noted for: data linkage (median, range of adoption: 45%, 30%-80%), access and accuracy of personal data (50%, 0%-100%) and anonymisation procedures (56%, 11%-100%). A high variability was noted in the application of privacy principles. CONCLUSIONS: A comprehensive methodology of Privacy and Ethics Impact and Performance Assessment was successfully applied at international level. The method can help implementing the GDPR and expanding the scope of Data Protection Impact Assessment, so that the public benefit of the secondary use of health data could be well balanced with the respect of personal privacy.

Data Rich, Information Poor: Can We Use Electronic Health Records to Create a Learning Healthcare System for Pharmaceuticals?

Judicious use of real-world data (RWD) is expected to make all steps in the development and use of pharmaceuticals more effective and efficient, including research and development, regulatory decision making, health technology assessment, pricing, and reimbursement decisions and treatment. A "learning healthcare system" based on electronic health records and other routinely collected data will be required to harness the full potential of RWD to complement evidence based on randomized controlled trials. We describe and illustrate with examples the growing demand for a learning healthcare system; we contrast the exigencies of an efficient pharmaceutical ecosystem in the future with current deficiencies highlighted in recently published Organisation for Economic Co-operation and Development (OECD) reports; and we reflect on the steps necessary to enable the transition from healthcare data to actionable information. A coordinated effort from all stakeholders and international cooperation will be required to increase the speed of implementation of the learning healthcare system, to everybody's benefit.

Development of a population-based microsimulation mode of physical activity in Canada.

BACKGROUND: Computer simulation modeling makes it possible to project physical activity levels and the prevalence of related health outcomes. Such projections can help to inform programs that aim to increase physical activity levels and improve population health. DATA AND METHODS: The Population Health Model (POHEM) platform was used to develop a dynamic microsimulation model of physical activity among Canadian adults. Key parameters were derived from the National Population Health Survey (1994/1995 to 2006/2007) and the 2000/2001 Canadian Community Health Survey. To assess the validity of the physical activity module (POHEM-PA), estimates from the simulation projections were compared with results from nationally representative surveys. RESULTS: Trends over time in physical activity levels, chronic disease prevalence, and Health Utilities Index based on POHEM-PA projections were similar to those based on data from subsequent cycles of the Canadian Community Health Survey. INTERPRETATION: The addition of a physical activity module to POHEM provides a tool that can improve understanding of the complex dynamics underlying the relationship between physical activity and health outcomes as a population ages.

International comparisons of health system performance among OECD countries: opportunities and data privacy protection challenges.

OBJECTIVE: Health data constitute a significant resource in most OECD countries that could be used to improve health system performance. Well-intended policies to allay concerns about breaches of confidentiality and to reduce potential misuse of personal health information may be limiting data use. A survey of 20 OECD countries explored the extent to which countries have developed and use personal health data and the reasons why data use may be problematic in some. RESULTS: Countries are divided, with one-half engaged regularly in national data linkage studies to monitor health care quality. Country variation is linked to risk management in granting an exemption to patient consent requirements; in sharing identifiable data among government authorities; and in project approvals and granting access to data. The resources required to comply with data protection requirements is a secondary problem. The sharing of person-level data across borders for international comparisons is rarely reported and there were few examples of studies of health system performance. DISCUSSION: Laws and policies enabling data sharing and data linkage are needed to strengthen national information infrastructure. To develop international studies comparing health care quality and health system performance, actions are needed to address heterogeneity in data protection practices.

Canadian Cancer Risk Management Model: evaluation of cancer control.

OBJECTIVES: The aim of this study was to develop a decision support tool to assess the potential benefits and costs of new healthcare interventions. METHODS: The Canadian Partnership Against Cancer (CPAC) commissioned the development of a Cancer Risk Management Model (CRMM)--a computer microsimulation model that simulates individual lives one at a time, from birth to death, taking account of Canadian demographic and labor force characteristics, risk factor exposures, and health histories. Information from all the simulated lives is combined to produce aggregate measures of health outcomes for the population or for particular subpopulations. RESULTS: The CRMM can project the population health and economic impacts of cancer control programs in Canada and the impacts of major risk factors, cancer prevention, and screening programs and new cancer treatments on population health and costs to the healthcare system. It estimates both the direct costs of medical care, as well as lost earnings and impacts on tax revenues. The lung and colorectal modules are available through the CPAC Web site (www.cancerview.ca/cancerrriskmanagement) to registered users where structured scenarios can be explored for their projected impacts. Advanced users will be able to specify new scenarios or change existing modules by varying input parameters or by accessing open source code. Model development is now being extended to cervical and breast cancers.

Informing policy makers about future health spending: a comparative analysis of forecasting methods in OECD countries.

OBJECTIVE: Concerns about health care expenditure growth and its long-term sustainability have risen to the top of the policy agenda in many OECD countries. As continued growth in spending places pressure on government budgets, health services provision and patients' personal finances, policy makers have launched forecasting projects to support policy planning. This comparative analysis reviewed 25 models that were developed for policy analysis in OECD countries by governments, research agencies, academics and international organisations. RESULTS: We observed that the policy questions that need to be addressed drive the choice of forecasting model and the model's specification. By considering both the level of aggregation of the units analysed and the level of detail of health expenditure to be projected, we identified three classes of models: micro, component-based, and macro. Virtually all models account for demographic shifts in the population, while two important influences on health expenditure growth that are the least understood include technological innovation and health-seeking behaviour. DISCUSSION: The landscape for health forecasting models is dynamic and evolving. Advances in computing technology and increases in data granularity are opening up new possibilities for the generation of system of models which become an on-going decision support tool capable of adapting to new questions as they arise.

Health behavior change following chronic illness in middle and later life.

OBJECTIVES: Understanding lifestyle improvements among individuals with chronic illness is vital for targeting interventions that can increase longevity and improve quality of life. METHODS: Data from the U.S. Health and Retirement Study were used to examine changes in smoking, alcohol use, and exercise 2-14 years after a diagnosis of heart disease, diabetes, cancer, stroke, or lung disease. RESULTS: Patterns of behavior change following diagnosis indicated that the vast majority of individuals diagnosed with a new chronic condition did not adopt healthier behaviors. Smoking cessation among those with heart disease was the largest observed change, but only 40% of smokers quit. There were no significant increases in exercise for any health condition. Changes in alcohol consumption were small, with significant declines in excessive drinking and increases in abstention for a few health conditions. Over the long term, individuals who made changes appeared to maintain those changes. Latent growth curve analyses up to 14 years after diagnosis showed no average long-term improvement in health behaviors. DISCUSSION: Results provide important new information on health behavior changes among those with chronic disease and suggest that intensive efforts are required to help initiate and maintain lifestyle improvements among this population.

Trajectories of health-related quality of life by socio-economic status in a nationally representative Canadian cohort.

BACKGROUND: Mortality and morbidity have been shown to follow a 'social gradient' in Canada and many other countries around the world. Comparatively little, however, is known about whether ageing amplifies, diminishes or sustains socio-economic inequalities in health. METHODS: Growth curve analysis of seven cycles of the Canadian National Population Health Survey (n=13,682) for adults aged 20 and older at baseline (1994/95). The outcome of interest is the Health Utilities Index Mark 3, a measure of health-related quality of life (HRQL). Models include the deceased so as not to present overly optimistic HRQL values. Socio-economic position is measured separately by household-size-adjusted income and highest level of education attained. RESULTS: HRQL is consistently highest for the most affluent and the most highly educated men and women, and is lower, in turn, for middle and lower income and education groups. HRQL declines with age for both men and women. The rate of the decline in HRQL, however, was related neither to income nor to education for men, suggesting stability in the social gradient in HRQL over time for men. There was a sharper decline in HRQL for upper-middle and highest-income groups for women than for the poorest women. CONCLUSION: HRQL is graded by both income and education in Canadian men and women. The grading of HRQL by social position appears to be 'set' in early adulthood and is stable through mid- and later life.

Validation of population-based disease simulation models: a review of concepts and methods.

BACKGROUND: Computer simulation models are used increasingly to support public health research and policy, but questions about their quality persist. The purpose of this article is to review the principles and methods for validation of population-based disease simulation models. METHODS: We developed a comprehensive framework for validating population-based chronic disease simulation models and used this framework in a review of published model validation guidelines. Based on the review, we formulated a set of recommendations for gathering evidence of model credibility. RESULTS: Evidence of model credibility derives from examining: 1) the process of model development, 2) the performance of a model, and 3) the quality of decisions based on the model. Many important issues in model validation are insufficiently addressed by current guidelines. These issues include a detailed evaluation of different data sources, graphical representation of models, computer programming, model calibration, between-model comparisons, sensitivity analysis, and predictive validity. The role of external data in model validation depends on the purpose of the model (e.g., decision analysis versus prediction). More research is needed on the methods of comparing the quality of decisions based on different models. CONCLUSION: As the role of simulation modeling in population health is increasing and models are becoming more complex, there is a need for further improvements in model validation methodology and common standards for evaluating model credibility.