Hospitals and health systems increasingly are adopting artificial intelligence technologies built into health IT such as EHRs, telehealth services, remote patient monitoring and other tools that can expand access to care and improve health equity.
At the same time, conversations around AI safety and regulatory measures are happening around the globe, including a major recent executive order issued by President Biden.
As AI becomes embedded in the daily lives of people across the country, learning how to best use the technology while prioritizing security and equity – both on the individual level and more broadly – is paramount.
Researchers at the University at Albany’s School of Public Health are actively exploring ways that artificial intelligence and machine learning can be applied to public health, to enhance health outcomes for patients while prioritizing patient safety and data security.
Xin Wang is an assistant professor in the department of epidemiology and biostatistics at the University at Albany. His areas of expertise include artificial intelligence, deep learning, precision medicine and medical image computing, among other related subjects.
Here he shares his thoughts on how public health researchers can support the development of safe and effective health-centered AI.
Q. Where do public health and artificial intelligence intersect? Where can AI help boost public health?
A. The intersection of public health and artificial intelligence represents a transformation in healthcare, offering unprecedented opportunities to enhance disease prevention, management and health promotion.
In the realm of disease surveillance, AI stands as a powerful tool. By using advanced algorithms such as deep learning techniques, AI can learn through large-scale datasets, including social media trends, healthcare records and environmental factors, to predict disease outbreaks and their potential spread.
This predictive capacity enables public health officials to implement targeted interventions, optimize resource allocation and formulate proactive strategies, thereby mitigating the impact of potential health crises.
AI can also be used for personalized health interventions. By analyzing individual health data, AI can tailor health education and preventive measures to individual needs, enhancing the effectiveness of public health activities.
This personalization extends to risk assessment, where AI algorithms can identify individuals at high risk for certain conditions, enabling early interventions and more effective management of diseases.
In public health research, AI can accelerate the steps of discovery and insights. Its ability to process and analyze complex and large-scale datasets transcends human capabilities, uncovering patterns and associations. This aspect is particularly important in epidemiological studies and in understanding the social determinants of health, where multifactorial influences converge.
AI-driven analytics can reveal insights into how socioeconomic factors, lifestyle choices and environmental exposures intersect to influence health outcomes.
AI also has the potential to enhance precision medicine in public health. By analyzing genetic data, AI can contribute to the development of personalized medicine strategies, leading to treatments and preventive measures that are specifically tailored to individual genetic profiles, thereby enhancing the efficacy and efficiency of healthcare delivery.
Q. How can CIOs and other health IT leaders help guarantee the security of AI deployments?
A. For health IT leaders, ensuring the security of AI deployments in healthcare is a multifaceted challenge that requires a comprehensive approach.
The first step lies in establishing robust data governance and privacy protocols. Healthcare data is inherently sensitive, and the integration of AI necessitates the handling of various quantities of personal health information.
To protect this data, health IT leaders must implement stringent data privacy measures aligned with regulations, such as HIPAA in the U.S. This involves encrypting data, regular security audits, and ensuring that AI algorithms are transparent and compliant with data protection laws.
Additionally, developing a clear data governance framework that outlines who has access to what data and for what purpose is important. Such frameworks can protect patient privacy, which is an essential component in healthcare settings.
The second crucial aspect is the establishment of a robust cybersecurity infrastructure. AI systems, by virtue of their complexity and interconnectedness, can be vulnerable to cyber threats like data breaches and AI-specific threats like model poisoning.
To counter these risks, health IT leaders should invest in state-of-the-art cybersecurity tools and infrastructure, including advanced intrusion detection systems and secure cloud services. Equally important is the development of a cybersecurity-aware culture within the organization.
Regular training and awareness programs for all staff, focusing on the best practices for data handling and the recognition of potential cyber threats, are essential. This human element is often the first line of defense against cyberattacks. Health IT leaders should also consider employing specialized AI security experts who can ensure that the organization’s defenses are always one step ahead.
Finally, a proactive, continuous monitoring and evaluation strategy is important for the security of AI deployments in healthcare. AI systems, by their nature, evolve and learn over time, which can introduce new vulnerabilities or unexpected behaviors.
Continuous monitoring of AI systems for any anomalies in performance or data handling is essential. This includes regular reassessment of risk management strategies and updating security protocols as AI systems and cyber threats evolve.
In addition, it’s important to establish an incident response plan that can be swiftly enacted in the event of a security breach. By adopting a proactive, vigilant approach to AI security, health IT leaders can not only protect sensitive data but also build a resilient, trustworthy AI system for healthcare.
Q. How can CIOs and other health IT leaders help guarantee AI deployments help boost health equity?
A. Ensuring AI contributes to health equity is a key responsibility for health IT leaders. The first step is the conscious and deliberate integration of equity considerations into AI development and deployment strategies. This begins with the diversity of data sets used to train AI algorithms.
Historically, healthcare data has often been skewed toward certain demographics, leading to biased AI systems that do not perform equitably across diverse populations. To counter this, health IT leaders must advocate for and facilitate the collection and incorporation of data from a wide range of demographic groups, including underrepresented minorities, to ensure AI tools are effective and fair for all patients.
Additionally, it is essential to involve a diverse group, including patients, clinicians and community representatives, in the development process to ensure the AI solutions address a broad spectrum of needs and perspectives.
The second key area of focus is the accessibility and usability of AI-driven tools. Health IT leaders must ensure AI solutions are designed with user-friendly interfaces accessible to people with varying levels of digital literacy and in different languages, as needed.
This includes the deployment of AI applications in a variety of healthcare settings, including under-resourced clinics and community health centers, to ensure equitable access. Moreover, health IT leaders should advocate for the integration of AI tools into telehealth services, which can significantly enhance access to healthcare for individuals in remote or underserved areas.
By prioritizing ease of use and widespread accessibility, they can help bridge the digital divide in healthcare, making advanced AI tools a resource for all rather than a privilege for a few.
Lastly, continuous evaluation and adaptation are important for ensuring health equity in AI deployments. This involves setting up mechanisms to regularly assess the impact of AI tools on different patient populations and identifying any disparities in outcomes or access.
Health IT leaders should collaborate with healthcare providers, patients and community organizations to gather feedback and insights on the performance of AI applications. Based on these evaluations, AI systems may need to be adjusted or retrained to address any identified disparities. This iterative process not only enhances the performance of AI tools but also demonstrates a commitment to continuous improvement in the pursuit of health equity.
Q. How can AI help enhance health outcomes for patients in the arena of public health?
A. AI holds tremendous potential to enhance health outcomes in public health. Firstly, AI can significantly improve disease surveillance and predictive analytics. By processing and analyzing vast amounts of data from various sources, AI can identify patterns and predict outbreaks of infectious diseases before they spread widely.
This early detection is important for timely intervention, allowing for more effective prevention strategies and resource allocation. AI can also be instrumental in disease management, a key area in public health. By analyzing patient data over time, AI algorithms can predict individual patient risks, suggest personalized treatment plans, and even alert healthcare providers to early signs of complications.
This proactive approach can lead to earlier interventions, better disease management, and, ultimately, improved health outcomes.
Secondly, AI can help bridge the gap in healthcare access by offering initial guidance, reminding patients about medications, and providing health education. For example, AI- and large language model-powered chatbots and virtual health assistants can provide basic healthcare guidance and information, especially in underserved areas where medical professionals are scarce.
This is particularly valuable in managing public health crises, where accurate information and guidance are important. Additionally, AI-driven telemedicine platforms can facilitate remote consultations and diagnostics, making healthcare more accessible to those in remote or underserved areas. This not only helps in managing existing conditions but also in preventive healthcare, as people can receive advice and screenings without the need for physical travel to healthcare facilities.
Lastly, AI can optimize healthcare delivery and resource allocation, which are important in public health settings. Through predictive modeling, AI can help public health administrators make informed decisions about where to allocate resources for maximum impact.
For example, AI can predict which areas are likely to face healthcare provider shortages or which communities are at higher risk for certain health issues, allowing for targeted interventions. This efficiency can lead to improved patient outcomes as healthcare providers can spend more time on direct patient care and less on administrative tasks.