Data scientists are creating machine learning models to predict infection and mortality rates and to determine resource needs and allocation based on these predictions.

Public health infrastructures civic and economic systems are being stressed beyond their boundaries due to the battle against COVID-19. To aid in the fight against the virus, data and AI technologies are being used in new ways, particularly in countries that adopt a scientific approach to public health. Data scientists are creating machine learning models to predict infection and mortality rates and to determine resource needs and allocation based on these predictions.

AI can be used to power two key tasks of pandemic mitigation:

· infection tracking

· infection spread prediction

If done correctly, AI can help uncover three foundational pieces of information, crucial to tracking and predicting the spread:

1. Measuring social isolation by observing individual mobility

2. Identifying clusters of more than a certain number of individuals and identifying the corresponding locations

3. Risk assessment of individuals and locations, at scale, by understanding the movement of infected individuals

With the absence of prevention and cure methods, the pandemic is primarily being managed through mitigation. One important aspect of mitigating virus spread has been through “contact tracing” which involves tracking the movements of infected people to identify and alert who came in contact with them. Contact tracing is a time-consuming process, but the unprecedented speed and scale of this pandemic makes these manual methods nearly impossible, not to mention less effective. Instead, data scientists are leveraging third-party data to create machine learning models to predict infection rates, which in turn can inform decisions about resource needs and allocation.

Data scientists are creating models to track the spread of the virus and helping resource allocation based on the prediction of hard-hit areas. AI, particularly, big data and machine learning techniques can be used to identify the infection risk of individuals, which can then be projected to those individuals and others in the geographic locations they have visited. AI is an enabler; it identifies patterns and provides insights at speeds well beyond what humans can do manually. But, the key to the successful use of AI relies on the data that is being fed into the models. If this data is inaccurate or lacks scale the ability of the model to predict outcomes will be impacted in a negative way. Data can be obtained in various ways, either by requesting information directly from individuals or by seeking data from other available sources.

Another, perhaps more reliable option, is to use other available data sources that can model the activities of the population at scale. In many cases location data and behavioral data can be used as inputs to COVID-19 predictive models. Mobilewalla provides data around individual mobility that acts as a proxy for social distancing. Mobilewalla can provide both a social isolation score, or the separate data attributes or features that can be used to build a custom score.

Mobilewalla is working with various businesses and municipalities providing data around individual mobility that acts as a proxy for social distancing. Mobilewalla is providing three types of related data for COVID-19 mitigation efforts:

1) Individual mobility, which is an indication of the distance someone travels on a regular basis,

2) Cluster identification which identifies gatherings of more than a certain number of devices (typically 10-20) at a single location, and

3) Building a contagion graph through individual device data, the radius of gyration and cluster identification.

Mobilewalla data is being used by health services organizations and governmental entities around the world to better predict the spread of the Novel Coronavirus at both the macro (city/county/state/country) and micro (predicting patients at a hospital) level. Mobilewalla can provide both a social isolation score, or the separate data attributes or features that can be used to build a custom score. This data includes individual mobility metrics (indicating the daily distance traveled and unique locations), cluster identification (gatherings of a high number of devices) and individual device data at both the micro and macro levels. These are all foundational inputs that can be used in COVID-19 prediction models.