Starts:

Tuesday, September 20th
10:45am-12:15pm EDT

Category:

Topical Workshop

Tracks:

Informatics, coding and pain registries

Room

716 A

It’s Difficult to Make Predictions: Data Mining and Machine Learning in Pain Research and Management

Prescribed and voluntary open data reporting practices, together with a growth in multinational research consortia, are generating medium to large clinical and experimental datasets of varying specificities and quality (in terms of pain information). These datasets provide avenues to explore complex relationships yet need innovative analysis approaches. In this workshop we will explore the application of data analytics to such datasets across a diverse range of pain-related topics, demonstrating the flexibility and power of analytical methods to generate novel insights in both basic science and clinical research. Jan Vollert will open the workshop by giving an introduction into data mining and machine learning, setting foundation knowledge about best practices and pitfalls in a fast-evolving, hot topic. Peter Kamerman will speak on using publicly accessible datasets (e.g., global burden of disease data, NHS England prescribing data, and general health survey data from the US) to gain high-level epidemiological insights into pain burden and the management of pain. Finally, Janne Gierthmühlen will present data on predicting development of pain as a result of diabetic polyneuropathy.

Presentations

Time
10:45am EDT12:15pm EDT

Extracting Useful Epidemiological Data on Pain from Public Datasets

Tracks: Informatics, Coding And Pain Registries
Categories: Topical Workshop

Numerous health-related open datasets exist, many of which span decades. Examples include the Global Burden of Disease dataset, NHS England GP prescribing data, and US health population surveys such as the National Health and Nutrition Examination Survey (NHANES). While these datasets are public, they either were not designed to directly address issues related to pain, or they are heterogenous (within and between surveys) in terms of the pain-related questions that have been asked over time. Nevertheless, these are rich datasets for obtaining high-level epidemiological data on pain, its consequences, and its treatment. In this presentation the wealth of information contained in these datasets and the methods used to identify, extract and analyse such data in order to achieve novel insights into pain at national and global levels will be shown to the audience.

10:45am EDT12:15pm EDT

Machine Learning: What it Can, Cannot, and Should Not Do

Tracks: Informatics, Coding And Pain Registries
Categories: Topical Workshop
Presented By: Jan Vollert

While the statistical models for machine learning have mostly been developed in the 20th century, with the wide availability of cheap computing and virtually endless data space, data mining and machine learning have become an accessible and mighty tool for researchers outside highly specialized groups. Beyond research, they have become a ubiquitous part of our daily life, as predictive text suggestions on our emails or face recognition to unlock our phones. This presentation will provide an overview of history, methods, and increasing usage of machine learning, from high-end research over daily life improvements to preclinical and clinical opportunities for pain research. A special focus will be set on common misconceptions, inherent bias, and the dangers of (mis)use of computational decision making.

10:45am EDT12:15pm EDT

What are Risk Factors for Painful Polyneuropathy? An Aalysis Using Machine Learning

Tracks: Informatics, Coding And Pain Registries
Categories: Topical Workshop

We are in an era where prescribed and voluntary open data reporting practices, together with a growth in multinational research consortia, are generating medium to large clinical and experimental datasets of varying specificities and quality (in terms of pain information). These datasets provide avenues to explore complex relationships yet need innovative analysis approaches. In this workshop we will explore the application of data analytics to such datasets across a diverse range of pain-related topics, demonstrating the flexibility and power of analytical methods to generate novel insights in both basic science and clinical research. Jan Vollert will open the workshop by giving an introduction into data mining and machine learning, setting foundation knowledge about best practices and pitfalls in a fast-evolving, hot topic. Peter Kamerman will speak on using publicly accessible datasets (e.g., global burden of disease data, NHS England prescribing data, and general health survey data from the US) to gain high-level epidemiological insights into pain burden and the management of pain. Finally, Janne Gierthmühlen will present data on predicting development of chronic pain in longitudinal studies of populations at risk, like diabetes.

Presenters

Jan Vollert

Research Fellow
Imperial College London

Professor Janne Gierthmühlen

Senior physician and researcher
Universitätsklinikum Schleswig-Holstein, Campus Kiel; Sektion neurologische Schmerzforschung und -therapie

Professor Peter Kamerman

Professor
University of the Witwatersrand