Harnessing the Power of AI during Epidemics

Even as the world gains, at certain moments at least, tentative handle on Covid-19, the pace of globalisation makes it likely that further pandemics are in the pipeline, states an article in Nature. When it comes to population-level health interventions, policymakers and governments are often called on to make difficult decisions involving complex trade-offs between public health and economic advancement.

One viewpoint is that when it comes to travel bans, closing businesses, mandating masks, testing, etc., we should let data-driven artificial intelligence (AI) models help guide public health recommendations. Here are a number of ways that AI can be harnessed during epidemics.

Forecasting disease dynamics

Machine learning, part of AI, makes use of algorithms that use data and experience for automatic updates and improvements. These AI models can predict short or longer-term outcomes, including - as the Nature article notes -“infections, deaths, and effects of non-pharmaceutical interventions”. This helps to inform decision making when it comes to pandemic management.

Real-time monitoring of adherence to public health recommendations

There are various ways AI models can be used to assess compliance with public health mandates or recommendations. For example, computer vision systems were developed to monitor interactions with Covid-19 patients in hospitals to determine - for instance - whether PPE was secure, which employees entered the room and the duration of contact with patients. But bear in mind that some of these interventions - e.g. AI-based facial recognition software, which was used in China during Covid-19 - raise serious privacy concerns.

Real-time detection of symptoms

AI can be used for early detection of illness, such as Covid-19, through data collected from wearable devices. For example, the Oura Ring, developed by the Rockefeller Neuroscience Institute, uses an AI-guided model to help predict the onset of Covid-19 symptoms 3 days in advance with 90% accuracy by tracking physiologic measures together with psychological, cognitive and behavioural biometrics.

Surveillance and outbreak detection

AI models can track potential exposures or symptoms in real time. This data can be integrated with precise geolocation information to make accurate predictions about where outbreaks are taking place.

Prognosis prediction

AI algorithms can help prioritise those patients likely to experience worse symptoms of a particular illness. Machine learning is particularly adept at making sense of large amounts of complex and / or unstructured data. A wide variety of potentially useful data, from clinical symptoms to known exposures in the area, can be fed into such models.

AI has emerged as a key area of investment for healthcare leaders, according to the Future Health Index 2021 report. This survey analyses insights from nearly 3,000 healthcare leaders across 14 countries. The report states that roughly one third (36%) of European healthcare leaders and one quarter (27%) of those in APAC agree that “to be prepared for the future, their hospital or healthcare facility most needs to invest in implementing predictive technologies like AI and machine learning”.

Healthcare, like so many other industries, is not immune to disruptive change. However, these changes should, on the whole, be embraced. By harnessing the power of AI during epidemics, so many goals can be achieved, including better utilisation of resources, disease prediction, locating outbreaks and assisting patients optimally. Ultimately, these technological advancements have enormous benefits for individuals and society at large.