SOCKS (Statistical Outlier Curation Kernel Software): A Modern, Efficient Tool for Outlier Detection and Curation

Overview

Real-world signal acquisition through sensors is at the heart of the modern digital world, yet almost every signal acquisition system is contaminated with noise and outliers. Our data “sanitizer” tool provides a way to obtain uncorrupted observations.

Market Opportunity

It is increasingly evident that clarity of digital information is more important than volume of data. Data are often contaminated by background noise and outliers, which blur the underlying nature of the observed system. With the ever-growing volume of digital data, there is a critical need for a generic yet customizable outlier-detection and curation tool. This is especially important in real-time biomedical applications such as EKG, which influence clinical decisions. Existing software tools merely provide data smoothing, which inevitably causes loss of valuable information. An effective open-source software tool for outlier detection and raw-data curation can help to solve the “garbage in, garbage out” challenge with traditional artificial intelligence and machine learning algorithms.

Innovation and Meaningful Advantages

Our novel SOCKS (Statistical Outlier Curation Kernel Software) tool decontaminates time-series and matrix-like data sources to recover the ground truth. Our approach represents a fundamental shift in the design of real-world data-processing tasks. We decontaminate raw data first, through conditional flagging of outliers, and then curate the flagged data points, enabling recovery of asymptotically converged ground truth. The process greatly increases the accuracy of traditional data-processing tasks. With this robust, easy-to-use, low-latency, generic yet custom-izable outlier-detection and curation tool, there are no practical size limitations of the input data source.

The SOCKS tool curates data on either time domain or spatial domain (or image), to produce filtered curated data while separately storing the outlier data points. The output can be further processed internally using traditional auxiliary filters. Because the flagging and curation steps occur sequentially, once the outliers are flagged, curation is optional and the flagged data records can simply be stored. Potential applications of the SOCKS tool would span nearly all data-driven systems with or without outliers (the definition of outlier often being arbitrary). Disciplines in which SOCKS would find immediate application are EEG, EKG, MRI, ECHO and various bio-imaging modalities, financial and other time-series, astrophotography, microscopic images, internet network-traffic monitoring, and fraud-detection systems.

Collaboration Opportunity

We are interested in exploring 1) startup opportunities with investors; 2) research collaborations with leading software development companies; and 3) licensing opportunities with companies.

Principal Investigator

Prasanta Pal, PhD

Investigator in Epidemiology

Brown University

prasantapal@palinnovations.tech

IP Information

US Utility Filed, Priority Date August 12, 2021

Publication

Pal P, Van Lutterveld R, Quirós N, Taylor V, Brewer, Judson B. Statistical Outlier Curation Kernel Software (SOCKS): A Modern, Efficient Outlier Detection and Curation Suite. TechRxiv. Preprint. Posted 2021 July 28. doi.org/10.36227/techrxiv.15043695.v2.

Contact

Melissa Simon, PhD

Director of Business Development

melissa_j_simon@brown.edu

Brown Tech ID 3108

Website

http://brown.technologypublisher.com/technology/48743

Contact Information

TTO Home Page: http://brown.technologypublisher.com

Name: Melissa Simon

Title: Director of Business Development

Department: BTI

Email: melissa_j_simon@brown.edu