The human brain cannot possibly integrate the vast and rapidly growing amount of information modern societies have been able to amass. Our knowledge network (SPOKE) integrates hundreds of millions of biomedical concepts into a knowledge engine to enable doctors, drug developers and citizen scientists connect the dots and produce a biologically meaningful answer to these questions. 

Over the course of the project period, SPOKE has grown in size by a factor of 15, with 40 million nodes representing various biomedical & health related concepts and 100 million edges representing connections among those concepts. More than 50 independent data sources have now been integrated into SPOKE by the team, increasing capabilities for network exploration in the areas of molecular pathways as well as the environmental factors that may affect health.

During the award period, the team expanded development of a simple knowledge network visualizer (the SPOKE neighborhood explorer) to support interactive exploration. Most importantly, a commercial entity was spun off from the core SPOKE team in partnership with UCSF Innovation Ventures.  Mate Bioservices develops a comprehensive network interface with customizable analysis tools to meet the needs of a variety of user profiles including biomedical scientists, drug developers, clinicians, and citizen scientists. The minimum viable product for this interface, BRIC (Biomedical Researchers’ Intellectual Companion), will be made available by Q1 2024.   

Recognizing that there are significant ethical, legal, and social implications associated with the development and use of such powerful tools for biomedical knowledge exploration, the team has generated a risk assessment report to address usage risk, responsibilities to mitigate negative impacts, potential mosaic effect, validity, trustworthiness and generalizability. 

As part of its academic development, the team examined the baseline global features of SPOKE and continues to monitor these features to ensure that structural artifacts do not emerge. The SPOKE Team also partnered with Oak Ridge National Laboratory to achieve exaflops computing, a work recognized by reaching the finalist stage of the Gordon Bell Prize competition. The team has also demonstrated the utility of the network and the validity of a network-based analysis approach to generate hypotheses and provide prognostic information about patients. In March of 2020 SPOKE integrated the newly published SARS-CoV-2 Interactome into the network to explore some potential pathways of viral activity and were able to identify pathways and predict promising therapeutic measures that were later confirmed experimentally in the literature. 

We have also shown the utility of the network when observed data is overlayed, and signature patterns emerge as feature weights are modeled across the network. For example, our team recently published work on an algorithm to embed millions of deidentified electronic health records (EHRs) into SPOKE to produce high-dimensional, knowledge-guided patient health signatures that were subsequently used as features in a machine learning environment to classify patients at risk of developing a chronic disease. Our model predicted disease status 3 years before being diagnosed with multiple sclerosis (MS). The method outperformed predictions using EHRs alone, and the biological drivers of the classifiers provided insight into the underpinnings of prodromal MS. 

As a notable outcome of this award, we recently incorporated Mate Bioservices, a company that commercializes applications of SPOKE. Mate Bioservices secured the exclusive license for SPOKE from UCSF, and filed a patent for SPOKE and our proprietary algorithms. Mate has already retained a Chief Executive Officer, who is actively seeking customers and investment opportunities from venture capital as well the NSF SBIR funding mechanism. Mate already has revenue from contracts with industry and academia and the team is currently in negotiations with potential investors to raise capital for expansion

In summary, throughout our participation in the NSF Convergence Accelerator program, we have developed and made available a fully functional biomedical KN. We have established a Governance Committee for SPOKE, guiding expansion of the KN with additional knowledge sources; and developed an open access network visualization tool. We have completed a number of analyses and published results validating the network’s quality and utility. We secured rights for commercialization of SPOKE-powered products, and established a company to support commercialization of four products powered by the KN.

Last Modified: 12/16/2023
Modified by: Sergio E Baranzini