Forecasting emerging technologies

Swedish information scientists collaborate with global pharmaceutical company and data mining experts to forecast technologies related to intelligent pharmaceuticals.

Through high performance research & education networks researchers from the Universities of Skövde and Borås in Southern Sweden gain access to about 700,000 different sources on the open web, being analyzed to forecast emergent technology related to what is known as intelligent pharmaceuticals.

The research project is carried out in cooperation with global pharmaceutical company AstraZeneca and Swedish company Recorded Future, specialized in forecasting and analysis.

The project is named PET, “Picking the Winners: Forecasting Emergent Technology through Bibliometrics/Altmetrics, Topic Modeling and Information Fusion.”

“Predictive analytics is something that we, and many others, are looking into more closely these days,” says Nasrine Olson, senior lecturer and researcher from the Swedish School of Library and Information Science, University of Borås.

Probabilistic models

“With all the massive data available for mining, it has become possible to build probabilistic models that can indicate potential future trends. The PET project brings together techniques from research fields of information fusion, bibliometrics/altmetrics and topic modelling in order to further advance the state-of-the art within technology forecasting.

“The main use case for the project will be forecasting of technology related to intelligent pharmaceuticals, where the main concern is to develop technology for effectively supporting the health care providers and patients to obtain optimal effects of medications.

One step further

As Nasrine Olson points out, the conventional way to forecast emerging technologies has been by looking at scientific publications, which due to the time delay involved in the publication process, do not always represent the most recent research results. The PET project wants to go one step further. This means analyzing a vast amount of different sources, among other things conference programs, and even social media, to detect how different topics evolve over time.

“We’re doing text mining with focus of modelling how topics evolve in text corpora stemming from different sources on the open web, e.g., text flows from social media, news and databases with scientific publications related to intelligent pharmaceuticals.

“To achieve this, we’re using a wide range of methods from the areas of data analysis, machine learning, bibliometrics, information science, information analysis, information fusion, algorithms, uncertainty modeling and visualization.

“And we’re aiming at developing and applying advanced hierarchical models and algorithms to detect and predict how topic flows evolve over time based on a vast number of different sources on the open web.

The PET project is financed by Swedish research financier The Knowledge Foundation and will run until August 2018.