SBIR Banner

You are here

NIH/NCI 425 - Information Technology Tools for Automated Analysis of Physical Activity, Performance, and Behavior from Images for Improved Cancer Health

Fast-Track proposals will be accepted.

Direct-to-Phase II proposals will NOT be accepted.

Number of anticipated awards: 3-5

Budget (total costs, per award):

Phase I: up to $400,000 for up to 9 months

Phase II: up to $2,000,000 for up to 2 years




The exponential rise in availability of digital still and video imagery has created enormous opportunities for health researchers. However, software tools for automated image analysis in health are lacking. The goal of this topic is to stimulate development of software for automated analysis of physical activity, performance, and behavior in still and video images for clinical, home monitoring and public health applications. Physical activity refers to movement and postures such as ’walking’, ‘sitting’ or ‘standing up'. Performance refers to quantitative measures of function such as walking speed or timing of a sit stand test. Behavior refers to identification of specific actions such as ‘taking a pill’ or ‘playing soccer’.

Existing software tools emphasize counting and tracking customers (e.g. TraxSales), monitoring transportation behavior (e.g. TRAF-SYS), and security concerns in the private and defense sectors (e.g. DARPA Minds Eye Program). Additionally, emerging research is attempting to develop automated tools to assess sports performance and human performance capture for entertainment applications. In contrast, health-oriented applications are poorly developed, limited to a few publicly available image management and annotation tools. While larger companies are entering this sector (e.g. Microsoft AZURE), they also lack focus on health applications. Finally, advances in machine learning and AI research further support the potential for new products in this area.


Project Goals

This SBIR contract topic is designed to attract proposals for new and innovative image analysis tools to extract information concerning physical activity, performance and behavior. Each of these interrelated elements of human action have distinct associations with health and health monitoring needs. Examples include but are not limited to: 1) Automated assessments of gait, walking speed, and other medically-relevant performance parameters in the clinic; 2) Enabling in-home monitoring of compliance with medication and physical therapy regimens; and 3) Improved evaluation of physical activity in transportation or park settings. Potential image sources include, but are not limited to: wearable cameras, stationary cameras, smart phones, social media, Photovoice projects, and archives of street images from Google Street View or Gigapan. Applicants will be asked to specify the use case for their project and identify the source of images. Images may be from pre-existing sources or may be collected as part of the project. Collaboration with relevant subject matter experts and computer vision specialists is required to insure use of best possible analytical approaches. The tools developed must provide solutions for protecting sensitive or personally identifiable information available in the images.

The long-term goal of the project is to develop software that can automatically extract data from images concerning people and their activities. Advances in security, loss prevention, assessment of human behavior in retail environments, and automated measurement of human performance in sport and animation domains along with growing capacity of computers to identify and count objects via advances in artificial intelligence and machine learning suggest that algorithms are available that could be applied to health questions. Data from these algorithms could help multiple aspects of cancer prevention and control from primary prevention such as improved evaluation of interventions to encourage physical activity, to enhanced epidemiological studies, to automation in monitoring of symptoms and response to treatment for disease affecting physical performance, to improved compliance with cancer treatments or physical rehabilitation regimens. This interplay could advance health research and lead to improved commercial products for diverse applications.

Proposals addressing biomedical images such as MRIs, microscopy, or DEXA will not be deemed responsive to the call.


Phase I Activities and Deliverables:

• Establish a project team including proven expertise in: image analysis, including recognizing human actions and event segmentation; algorithms for data extraction, e.g. machine learning or neural networks; image data storage and manipulation; secure transmission of health data (if needed); user interface development; and topic-specific expertise in the appropriate behavioral science and public health domains.

• Develop a precis of the proposed tool and carry out structured interviews or one or more focus groups aimed at defining specific subject matter needs.

• Create or identify an open access image data source. Examples include, but are not limited to, cell phone images, SenseCam data, the AMOS archive of webcam images, Photovoice collected image libraries, and security video.

• Develop a functional prototype system from planned Phase I characteristics that includes:

• Capacity to extract data from at least one image type involving human physical activity, performance or behavior,

• Capacity to combine both automatic and manual detection and counting of intended aspects of physical activity, performance and behavior via a graphical user interface on a desktop computer, laptop computer, and/or tablet.

• Conduct a usability study with at least 15 users not affiliated with the study team and in several distinct user groups

• Provide a report including a detailed description and/or technical documentation of the proposed tool including plans for managing large numbers of image files, specific data resources and file formats targeted, details of the algorithmic approaches to be used and an assessment of potential bias in training image data sets, and approaches to be used to assess performance of the software tools. Comparison with gold standard measures such as human data extraction is an important part of validating the approach.

• Describe hardware and any additional software required for use of the tool.

• Present phase I findings and demonstrate the functional prototype system to an NCI evaluation panel via webinar.


Phase II Activities and Deliverables:

• Describe and document protocols and guidance for investigators working with imaging to insure appropriate informed consent, risk assessment, and data management.

• Present Phase II findings and demonstrate the software system to an NCI evaluation panel via webinar

• Improve and expand the capacity of the software to identify aspects of physical activity, performance, and behavior

• Develop or refine data extraction algorithms

• Further test reliability and validity of data extraction via new methods or new image file sources;

• Create a library of open access test images for additional algorithm training efforts;

• Propose and implement a cycle of usability testing incorporating user center design principles to enhance software ease and efficiency of use;

• Develop systems documentation where applicable to support the software and bioinformatic methods.

• In the first year of the contract, provide the program and contract officers with a letter(s) of commercial interest.

• In the second year of the contract, provide the program and contract officers with a letter(s) of commercial commitment.



Receipt date: October 26, 2020, 5:00 p.m. Eastern Daylight Time

Apply for this topic on the Contract Proposal Submission (eCPS) website.

For full PHS2021-1 Contract Solicitation, CLICK HERE




Posted Date: 
July 29, 2020