Introduction

Imagine a time when a computer programme or a phone application gives you an alert, predicts an upcoming flu outbreak in your community, or even has your routine medications smartly delivered by a drone during healthcare personnel shortages. This is not the plot of a sci-fi movie- these are only a few of the endless and yet unravelling potential of Artificial Intelligence (AI).

Artificial Intelligence, simply called AI, is the simulation of human intelligence in machines made to learn and think like humans.¹ It can be as simple as Amazon Alexa which responds to your commands or as complex as systems that can process vast amounts of data and learn from their experiences. 

COVID-19 was one of the most significant and life-changing pandemics in human history, opening up the frontiers to many discussions and attempts to integrate AI into healthcare. Therefore, robots, for instance, were deployed to carry out tasks such as hospital sanitation for infection prevention and control, delivering drugs and meals in patient’s wards; and the COVID Dialogue which helped people to have medical consultations virtually.² 

In addition, AI has been employed in various aspects of communicable disease surveillance and monitoring such as Monkeypox, Salmonella, and Dengue.³

Healthcare crises and pandemics are therefore critical periods in a country’s health system and economic dynamics hence, advanced technology like AI can be employed in the efficient and effective management of such vulnerable situations. 

These applications include but are not limited to: 

• Making predictions about outbreaks or crisis 
• Making timely diagnosis 
• Administering appropriate treatments 
• Proper communication with patients 
• Managing scarce resources during health emergencies

AI in prediction and early detection

Predicting disease outbreaks and transmission patterns

Healthcare data is big, complex, and difficult to manage with traditional statistical models alone. Machine learning (which involves teaching computers to learn from data) can combine different sources of information and complex data much better than the common statistical models widely used.¹ This helps improve accuracy, and boost performance as data from pandemics are often irregularly collected and with many dimensions to it. 

Additionally, it has been predicted that AI has the potential to deal with imperfect information and cover for data gaps during high-pressure emergencies. This is based on AI’s performance during and following the COVID-19 pandemic.^1,4

AI models predicting outbreaks

Currently, the use of machine learning has involved detecting the different flu viral strains transmitted by animals to humans with precision.³ This has enabled the prediction and modelling of different outcomes should a pandemic result such as the number of infections, and deaths, and then testing various interventions to help reduce these. 

Additionally, to improve public health monitoring, AI has allowed for increased data streams to help analyse and verify health threats.⁵

AI-driven diagnostic tools that detect diseases from medical imaging or genetic sequencing

One of the most profound applications of AI in disease outbreaks is its ability to distinguish between the different forms or strains of the COVID-19 virus.¹ This has allowed scientists to rapidly detect mutations in the virus, raising alarm that has helped global health to better conduct contact tracing by determining the genetic origin of each case of the disease. This has also served as one of the bedrocks for vaccine development.

Role of AI in monitoring and reporting real-time data to healthcare authorities

Nowadays, AI-enhanced real-time disease surveillance systems can use digital sources such as social media, sensors, etc., to identify communities and locations at risk of disease outbreaks.⁵ 

Some AI models such as the Stacked Long Short-Term Memory (SLSTM) have been observed to assist in forecasting outbreaks beyond merely identifying early symptoms and providing treatments to patients.³ 

AI for Science (AI4S) in particular has proven to be superior to traditional tracking methods by improving the precision by which sophisticated data is analysed to identify any increase in the frequency of disease cases thereby innovatively alerting health authorities about potential outbreaks.⁶

Similarly, AI helps to cover for health workforce shortages which is usually the case during outbreaks, as well as augmenting weak public health surveillance systems.⁵ This is particularly beneficial, although not restricted to, low and middle-income countries which generally have higher risks of cases and deaths.

AI in response and containment

Simulation models powered by AI to predict the spread and impact of diseases

The traditional models of predicting disease transmissions include the Susceptible-Infected-Recovered (SIR) model which is compartmentalised.¹ 

AI models, such as the agent-based model, have been demonstrated to provide better integrated networks in predicting the patterns of transmission considering the dynamism (i.e., rapid spread)  of infectious diseases.

Decision support systems for public health responses and resource allocation

One of the major perks of AI in healthcare crises is its ability to influence decision-making. High-pressure situations like outbreaks require pragmatism in dealing with incomplete or imperfect information, constantly changing data and rapidly evolving situations. 

AI almost perfectly assists decision-makers in prioritising and balancing options and allocating critical resources effectively.⁴ This fosters collaboration among stakeholders and in delivering responsibilities more effectively and as timely as possible. 

Helpful decisions such as quarantine, isolation measures, social distancing and other infection prevention and control measures can be instituted earlier on, long before any eventualities.

AI in treatment and vaccine development

AI in drug repurposing and development

As earlier stated, AI and machine learning have applications in rapidly yet meticulously studying the different forms of infectious agents as well as any changes within the genetic makeup of these agents.^1,3 This informs drug and vaccine development. In addition, AI-driven platforms have the potential for managing and predicting individual patient responses to treatments.

AI in healthcare logistics

Resource management

Hospital resources and supply chains may be improved through AI predictive models and blockchain.⁴ Other healthcare logistics where AI is being explored are: 

• Managing patient flow 
• Bed allocation 
• Critical care resources

Remote patient monitoring, communication and telemedicine

One of the most important uses of AI is in streamlining communication regarding risk, transmission, treatment and control measures. Healthcare crises are often confusing for communities and further riddled with ample information sources, some of which may be false or even harmful.³ 

Therefore, the ability of AI to help governments and individuals filter out unofficial sources of public health information is beneficial and applicable during outbreaks. This improves crisis communication, allays fears, and helps individuals and communities take well-informed actions in taking charge of their health. Common AI language models such as ChatGPT have been projected to be useful in managing future pandemics.

Equally, AI provided several platforms that enabled patient monitoring by health providers during the COVID-19 pandemic, allowing for virtual consultations and treatment follow-up.²

Ethical, privacy, and implementation challenges

While AI continues to evolve and prove promising in managing health crises and outbreaks, ethical questions and implementation practicalities must be considered and discussed across communities and governments.^2,4,7 Some of these considerations are listed below:

• Potential for patient harm due to errors from the AI models themselves or inconsistencies in client/patient information
• Misuse of AI tools by both health providers and individuals
• Perpetuation of inequities; there are instances of underreporting or poor referral of  black people to secondary care due to a bias in the design of the models⁷
• Lack of transparency and accountability
• Data privacy and security
• Some of the AI projections by models may be impractical in the real-world context considering the obstacles (such as research funding) and complexities of health systems⁷

Summary

The future of AI in managing healthcare crises and pandemics involves leveraging advanced predictive analytics, diagnostic tools, and resource management systems to enhance disease outbreak prediction, containment strategies, and personalised treatment options. Ethical considerations, privacy concerns, and implementation challenges must be addressed to fully harness AI's potential to revolutionise emergency healthcare responses and improve global public health outcomes.