What We Do
An increasing number of data sources exist that – with appropriate analysis – can yield information to predict and detect adverse food events, including supply disruptions caused by natural disasters, political instability, or market fluctuations. However, significant barriers exist that prevent optimal use of this data including the sheer amount of data to be analyzed. To harness the potential of this data, FPDI develops research methods and information technology tools to collect, interpret, and analyze the data streams allowing a range of stakeholders to identify adverse food events as early as possible, ideally before events cause public health harm.
How We Work
FPDI develops predictive analytics technology solutions by first understanding the specific needs of end users, including how “big data” can address their questions and challenges. From there, FPDI develops detailed case studies and methodically reviews and assesses data sets to gain an in-depth understanding of available data across multiple systems and how those data meet end user needs. FPDI then fuses disparate data streams, develops custom algorithms, and applies innovative methods to create predictive analytics tools – all with end users in mind.
Our team has experience with natural language processing, machine learning, and the development of domain-specific coding frameworks to support robust data analysis. FPDI has experience applying predictive analytics methods to many different types of data, including open source media, import and trade data, and social media. Examples of this approach include:
- FIDES – a web application designed to fuse multiple streams of data to predict, monitor, and identify food system disruptions and adverse food events
- Disease Watch – a web application that ingests real-time open source media to identify and characterize global disease outbreaks or pre-cursor events (natural disasters, civil unrest) that could increase the likelihood of a disease outbreak
We have in-house experts and a network of global collaborators active in the field of predictive analytics for food protection. Additionally, while most of our tools were originally developed for the food protection context, the underlying methods and technologies have been leveraged and applied to predictive analytics in other domains including infectious disease and water security.