AI tools currently support clinical decision-making at the point of care and the identification of patient panels to improve population health by supporting key functions like risk stratification and predictive modeling. At the public health level, predictive analytics and other new AI-enabled data analysis methods help anticipate community needs and assess options to address social determinants of health and their concomitant health outcomes. AI can also facilitate the development of targeted population health interventions and policies.
Olawade et al.(2023)6
have identified multiple uses of AI in public health, highlighting its benefits for predictive modeling and surveillance to guide public health measures. The U.S. Centers for Disease Control and Prevention (CDC) has used natural language processing (NLP) methods to enhance COVID-19 vaccine safety monitoring and analyzed large amounts of non-traditional data sources to track and forecast the spread of the disease (CDC, 2023, July). The agency continues to explore new applications of AI and machine learning for public health to forecast trends in opioid overdose mortality, identify potential foodborne outbreak sources, and facilitate syndromic surveillance to monitor early signs of public health threats.
Olawade et al.(2023)6
also highlight AI’s contribution to the real-time examination of a variety of data sources, such as social media, electronic health records, genomics, medical images, and sensor data. For instance, infodemiology and infoveillance can enhance the analysis of Internet search behaviors, social media patterns, and publication trends within specific populations. This could facilitate the identification of emerging public health issues and the ongoing analysis of social determinants of health to help design population-specific interventions. Precision public health (PPH) methods could become game changers in identifying specific community subgroups to help design, test, and deliver health promotion and disease prevention interventions (Fisher & Rosella, 2022; Ling Ong et al., . 2024).3,5
At the healthcare level, using gradient-boosting decision trees can support population health planning by predicting the incidence of preventable hospitalizations among subgroups (Fisher & Rosella, .2022).3
The ability to design population-specific strategies using AI can enhance the implementation of initiatives to reduce health disparities. For instance, the ability to improve the prediction of outcomes based on socioeconomic, environmental, and lifestyle factors could enhance the risk stratification of interactions between social determinants of health and health outcomes (Ling Ong et al., 2024).5
Moreover, a better understanding of behavior patterns according to socioeconomic, environmental, and lifestyle factors improves opportunities to adapt health messages to a population’s education, culture, or preferences and opportunities to support chronic disease self-management skills (Leal Neto & Von Wyl, 2024).4
Nonetheless, when considering the use of AI in healthcare or public health, a series of risks, limitations, and ethical challenges must be considered. It is important to keep in mind that generative AI models must be trained with high-quality data, as poor-quality data can lead to inconsistent, unreliable, or biased outcomes (Bharel et al., 2024).1
In fact, bias can be introduced at any point during the development and deployment of AI algorithms, exacerbating health disparities and limiting the effectiveness of population health strategies (Fisher & Rosella, 2022; Ling Ong et al., 2024).3,5
Additionally, there are major privacy and stigmatization concerns associated with using big data from social media sites, blogs, and forums (Fisher & Rosella, 2022).3
Using big data could also involve the unauthorized use of personal data in predictive analyses (Ling Ong et al., 2024).5
Access to technology has similarly been signaled as a significant concern among researchers and practitioners, as this can potentially worsen health disparities in underprivileged communities. Several researchers have identified digital inclusivity as a social determinant of health (Ling Ong et al., 2024).5
; (Leal Neto & Von Wyl, 2024).4
The digital divide could create barriers to care for specific subgroups with lower digital literacy or lack of access to state-of-the-art technologies or even basic broadband, especially the economically disadvantaged and geographically isolated (Ling Ong et al., 2024; (Leal Neto & Von Wyl, 2024; Fisher & Rosella, 2022).3,4,5
Health organizations must thoughtfully strategize AI implementation to harness AI’s potential and maximize its impact on health outcomes. Fisher & Rosella (2022)3
proposed six key priorities for the successful deployment of AI technologies by public health organizations:
- understanding and complying with contemporary data governance;
- investing in modernized data and analytic infrastructure and procedures;
- building workforce stills and recruiting staff with desired skills;
- developing strategic collaborative partnerships to gain expertise, share access to data and infrastructure, and engage different perspectives;
- applying good AI practices for transparency and reproducibility, following practical guidelines for AI development, evaluation, implementation; and
- assessing and implementing measures to reduce bias, including the validation of the AI model across subgroups and using an ethical AI framework such as the World Health Organization’s Framework on the Ethical Use of AI in Health.
Health organizations must lead in ensuring that AI models are grounded with high-quality, complete data. Organizations must also consider the potential operational and service delivery implications of AI tools, starting with assessing their capacity to effectively and ethically implement these technologies. It is also important to engage and work in partnership with subject-matter experts, including public health professionals, AI researchers, and people with lived experience within target communities. As we continue to develop and deploy AI-enabled strategies in public health and healthcare, it is imperative to emphasize the protection of human safety and well-being and ensure equitable opportunities for optimal health.
References
- Bharel, M., Auerbach, J., Nguyen, V., & DeSalvo, K. B. (2024). Transforming Public Health Practice With Generative Artificial Intelligence: Article examines how generative artificial intelligence could be used to transform public health practice in the US. Health Affairs, 43(6), 776-782.
- Centers for Disease Control and Prevention. (2023, July). Artificial Intelligence and Machine Learning: Applying Advanced Tools for Public Health. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. www.cdc.gov
- Fisher, S., & Rosella, L. C. (2022). Priorities for successful use of artificial intelligence by public health organizations: a literature review. BMC Public Health, 22(1), 2146.
- Leal-Neto, O., & Von Wyl, V. (2024). Digital Transformation of Public Health for Noncommunicable Diseases: Narrative Viewpoint of Challenges and Opportunities. JMIR Public Health and Surveillance, 10(1), e49575.
- Ling Ong, J. C., Seng, B. J. J., Law, J. Z. F., Low, L. L., Kwa, A. L. H., Giacomini, K. M., & Ting, D. S. W. (2024). Artificial intelligence, ChatGPT, and other large language models for social determinants of health: Current state and future directions. Cell Reports Medicine, 5(1).
- Olawade, D. B., Wada, O. J., David-Olawade, A. C., Kunonga, E., Abaire, O., & Ling, J. (2023). Using artificial intelligence to improve public health: a narrative review. Frontiers in Public Health, 11, 1196397.
