A biological marker (biomarker) can be defined by any substance, structure, or process of the body, its actions or products that can influence or predict the incidence of disease, health conditions, effects of treatments, and interventions. Digital biomarkers are the user generated physiological and behavioral measures collected via connected digital devices or wearable and mobile computing systems that can be used to explain, influence or predict the health related outcomes. The digital biomarkers do not include genetic information or data collected through traditional medical instruments. Examples of digital biomarkers include everything from geo-location and physical activity traces through internal physiological processes like vital signs to chemical makeups of skin, blood and other tissues collected by IoT devices, smartphones, wearables or novel digital computational platforms.
A set accurately and reproducibly measurable digital biomarkers can be used to predict various health conditions, outcomes. These can also be used to generate actionable insights. The rich data collected with the built-in sensors and processing units of our smartphones and wearable devices have already shown a lot of promise for passive and continuous measurement of several digital biomarkers which can be used to develop all sorts of new health sensing apps and interventions. However, there are a few major bottlenecks in digital biomarker computation, which include (i) the high level of initial setup cost and difficulty for a long-term data collection from a large population, (ii) the l in our existing smartphones, wearables or other digital tools capturing internal states and processes of the human body, (iii) various confounding factors compromising the accuracy and reproducibility of the digital biomarker, (iv) high cost and intrusiveness. The workshop aims to identify a set of solutions to these problems by bringing in key technological innovations from the areas of mobile computing, machine learning, health sciences and medicine.
The 1st Workshop on Digital Biomarkers, collocated with MobiSys 2017 offers a unified forum that brings academics, industry researchers and medical practitioners together and seeks novel, innovative and exciting submissions broadly related to the modeling, testing, and validation of new digital biomarkers for predicting incidence of diseases, health conditions, effects of treatments, and interventions. The workshop aims to facilitate a systematic discussion among experts from different knowledge domains including mobile sensing, systems, machine learning, medicine and health sciences in order to (i) identify new digital biomarkers capturing behavioral health, chronic and degenerative diseases, (ii) identify the key shortcomings of the existing mobile and wearable sensor systems, and research study software platforms (e.g., ResearchKit and ResearchStack) for digital biomarker inference in terms of scalability, customizability, and sensing affordances, (iii) find realistic solutions towards building new digital biomarker evidence engine leveraging sensor data from a variety of mobile systems (e.g., smartphones, wearables, IoT devices, or any novel sensor systems), (iv) identify key data collection, labeling, testing and validation methodologies of the new biomarker evidence engine.
Topics of interest (NOT an exhaustive list):
Predicting the incidence of disease, health conditions, effects of treatments, and interventions with digital biomarkers.
Design and implementation of mobile phone, wearable and/or novel embedded systems based computational platforms.
Integration of multimodal data from different sensor streams for digital biomarker modeling.
Using existing IoT infrastructure for new digital biomarker modeling.
Improved data collection, labeling, testing and Validation methodologies for digital biomarker modeling.
Novel signal processing or machine learning techniques for digital biomarker modeling.
Developing robust biomarker models that can handle data sparsity and mis-labeling issues.
Energy and resource efficient implementation of biomarker models.
Designing and implementing data feedback and visualization for both participants and caregivers.
Development of smartphone based automated health interventions with digital biomarkers.
Submission deadline: April 20 2017
Notification deadline: April 26 2017
Camera-ready due: May 1 2017
Workshop Date: June 23 2017
All submissions must be original work not under review at any other workshop, conference, or journal. Submissions can be 3 to 6 pages in length including references and must be a PDF file and anonymized. Submissions must follow the ACM MobiSys 2017 formatting guidelines.
The workshop accepts papers describing complete as well as novel work-in-progress studies. Submission link is here.
Tanzeem Choudhury, Cornell University and Health Rhythms
Andrew Campbell, Dartmouth College
Anind Dey, Carnegie Mellon University
Santosh Kumar, University of Memphis
Nic Lane, University College London and Bell Labs
Mirco Musolesi, University College London
Mayank Goel, Carnegie Mellon University
Mary Czerwinski, Microsoft Research
Cecilia Mascolo, University of Cambridge
Miah Wander, Microsoft Research
Graeme Rimmer, Google
Hane Aung, Cornell University
Fahim Kawsar, Bell Laboratories
Mashfiqui Rabbi, University of Michigan
Webmaster:Vincent Tseng, Cornell University