ORIGINAL UCSF MS BIOSCREEN 

In 2013, our UCSF Multiple Sclerosis (MS) Group, led by MS researcher, Dr. Stephen Hauser, and Pierre-Antoine Gourraud, announced the first prototype of the “Multiple Sclerosis BioScreen” - a milestone in the delivery of precision medicine for a chronic disease. This BioScreen, initially developed as an iPad-based tool, gained its full scope in the following years thanks to the work from the product & tech team led by Antoine Lizee. The MS Bioscreen is a data infrastructure platform that gathers all relevant MS data from different sources, including clinical, imaging, and biomarker information, visually represents the disease course of an individual with MS from a front-end interface, and frames this course within the context of a large cohort of patients treated according to contemporary standards. The goal is to inform more precise clinical decisions, and empower patients to participate more actively in their clinical care (Click here to watch a demonstration). This work received key support from PCORI and the Conrad N. Hilton Foundation.

 

THREE NEW DIRECTIONS FOR THE UCSF MS BIOSCREEN  

In 2015, Dr. Riley Bove joined the BioScreen project as its clinical lead, and expanded this proof-of-concept in several directions, to make versions of the tool that were more accessible to patients and their communities, and more convenient and actionable for doctors. To make this possible, our developer Erica Schleimer, joined the team in 2017.  

1. OPEN MS BIOSCREEN: AN OPEN PLATFORM FOR THE COMMUNITY 

This open-access, web-based version of the MS BioScreen launched in January of 2018!  

It is available to any patient, caregiver or clinician with a web browser. This will allow users within or beyond highly specialized academic care settings, the opportunity to enter data on their condition, obtain a richly contextualized, digestible and actionable predictive output, free of commercial interest, and participate in a shared decision-making process. This UCSF-based creation will spread the technological innovations of digital health, and be used to advance healthcare worldwide by its ability to map the highly individualized disease that is MS and capture its respective variation, globally.  

This project is generously supported by the Hilton Foundation

Will Rowles serves as liaisons to patients, providers, and the community. 

2. WIN BIOSCREEN: DIGITAL CLINICS OF THE FUTURE 

To bring forward state-of-the-art research grade information about our patients into our clinic visits and for our patients, doctors, and researchers to be on the same page, we have partnered with scientists from other brain diseases (the Autism group and the Memory and Aging Center) to build the WIN BioScreen. This platform pulls data from many disparate sources -- including the traditional electronic medical record, research studies (MRIs, genetics) and patient surveys -- then process and visualize the data in a single cohesive display. This application will be customized by the user so that the user can see all relevant data in one place to facilitate clinical and research decision making. 

Investigators: Riley BoveKate Rankin, Stephan Sanders

Developers: Erica Schleimer (project manager), Paul Sukhanov, Sindy Law, Michael Gilson, Adam Santaniello, Michael Schaffer, Jacob Spector. 

3. NeuroSHARE: DEPLOYING A PRECISION MEDICINE TOOL THROUGHOUT A HEALTH SYSTEM 

 

With generous support from Governor Brown’s California Initiative to Advance Precision Medicine (CIAPM), we have partnered with Sutter Health’s Research and Development Team to create NeuroShare. 

MS-Share, an interactive app, will instantly combine the latest precision medicine data from our UCSF group, with real-time data from the patient’s electronic health record, and with information that patients self-report about their symptoms between medical appointments. NeuroSHARE is being implemented in multiple Sutter general neurology practices. 

This partnership holds the promise of bringing precision medicine – precise treatment decisions to address needs of individual patients– directly to the diverse populations living with MS in Northern California.  

 

CREATING DATA-DRIVEN ALGORITHMIC TOOLS 

Combining the wealth of high-quality data with programmatic development and medical expertise, our team is building the algorithms that will power the future of digital health tools at the point of care.  We started with our patented 'Personalized Contextualization of Patient Trajectories'. Through his dissertation work, Antoine Lizee, a major force behind the BioScreen development, is currently validating machine-learning approaches to prediction of short-term outcomes. Data scientist Tanya Krishnakumar has joined the team to develop approaches for understanding how newly approved therapies might perform against existing therapies, in the “real world”. 

 

DIFFERENT SETS OF BACK-END PIPELINES  

Several innovations are improving the flow of quality clinical, MRI and other data through the MS BioScreen back-end architecture. Optimizations are constantly being implemented, by the back-end team, to ensure the most up-to-date data can be displayed smoothly on the MS Bioscreen. 

1. BUILDING A CLOUD-BASED MRI INFRASTRUCTURE 

A high-throughput imaging infrastructure is being developed to move MS exams acquired on UCSF MRI scanners or available in clinical PACS, through an automated de-identification and processing pipeline and into the cloud where they can be accessed by MS BioScreen client applications. The processing pipeline provides brain extraction, registration, and intensity normalization which in concert permit the direct comparison of serial data. This infrastructure lays the groundwork for developing advanced analysis pipelines to derive scalar imaging attributes that, together with clinical and other biomarker data in the MS BioScreen, can be mined and used to advance MS research.  

Key personnel: Jason Crane, Beck Olson 

2. MACHINE LEARNING IMAGE SIMILARITY CLASSIFICATION MODELING 

Visual and quantitative analysis of longitudinal changes in clinical MS imaging data (brain and spinal cord) currently requires specialized software to spatially align images between time points, and manual user input to select the images acquired with the same contrast mechanism (e.g. T1, T2, PD) for quantitative comparison. To do this at scale requires computational models capable of automatically identifying MR images at each exam time point acquired with similar tissue contrast for comparison. Machine learning methods are being utilized to develop an image similarity classifier that will enable automated computational analysis of longitudinal MS imaging data at scale.    

This work is supported by a grant from an anonymous donor.

Key personnel: Jason Crane (PI)  

3. EXTRACTING PRECISE INFORMATION FROM ELECTRONIC MEDICAL RECORDS 

The MS BioScreen does not only include information from our research participants, but also from our 7,000+ MS patients clinically followed at UCSF.  

Patient data (demographic, labs, medications, MS measures) from the UCSF Electronic Medical Records system have, under strict ethical guidelines, been extracted and uploaded to the MS BioScreen. The algorithms created for this project have also been made freely available for the research community. 

Postdoctoral fellow Dr. Vincent Damotte published this paper on his work. Tanya Krishnakumar will now ensure follow through for the next phases of the EMR project. 

 

DATA MANAGEMENT. 

None of these tools would be possible without a robust data management team, led by Adam SantanielloEPIC, the marquee longitudinal observational MS study at UCSF, provides a diverse array of datasets that have enriched and provided a template for several projects in the BioScreen ecosystem. Since 2008, the EPIC data infrastructure has been transformed from a small Access database into a web-based research platform for multiple studies that integrates clinical data, genomics, microbiome and blood bio-banking, cerebrospinal fluid analysis, optical coherence tomography, online patient assessments, and other data sources. The next generation of this data management platform is under development and features a focus on APIs to improve data source interconnectivity and expedite data access for the growing analytics talent pool in the MS group. Underpinning the EPIC and BioScreen software products and analytics efforts is a fleet of Linux servers and NAS devices, managed by the data team and housed at the Weill Institute and QB3. With the Weill and Open MS BioScreen initiatives forging a path, the incorporation of secure cloud computing resources to our data management toolkit is slated for 2018. 

Key personnel: Adam Santaniello, Adam Renschen.  

 

PAST CONTRIBUTORS

Antoine Lizee played a key role in making the MSBioscreen a major player in the digital health field. He initiated and led the development of the product vision, algorithmic core, and backend infrastructure from early 2013 until his departure to Europe to start a healthcare company in 2016.